Downloads & Free Reading Options - Results

We present NN-grams, a novel, hybrid language model integrating n-grams and neural networks (NN) for speech recognition. The model takes as input both word histories as well as n-gram counts. Thus, it combines the memorization capacity and scalability of an n-gram model with the generalization ability of neural networks. We report experiments where the model is trained on 26B words. NN-grams are efficient at run-time since they do not include an output soft-max layer. The model is trained using noise contrastive estimation (NCE), an approach that transforms the estimation problem of neural networks into one of binary classification between data samples and noise samples. We present results with noise samples derived from either an n-gram distribution or from speech recognition lattices. NN-grams outperforms an n-gram model on an Italian speech recognition dictation task.

We propose BlackOut, an approximation algorithm to efficiently train massive recurrent neural network language models (RNNLMs) with million word vocabularies. BlackOut is motivated by using a discriminative loss, and we describe a new sampling strategy which significantly reduces computation while improving stability, sample efficiency, and rate of convergence. One way to understand BlackOut is to view it as an extension of the DropOut strategy to the output layer, wherein we use a discriminative training loss and a weighted sampling scheme. We also establish close connections between BlackOut, importance sampling, and noise contrastive estimation (NCE). Our experiments, on the recently released one billion word language modeling benchmark, demonstrate scalability and accuracy of BlackOut; we outperform the state-of-the art, and achieve the lowest perplexity scores on this dataset. Moreover, unlike other established methods which typically require GPUs or CPU clusters, we show that a carefully implemented version of BlackOut requires only 1-10 days on a single machine to train a RNNLM with a million word vocabulary and billions of parameters on one billion words. Although we describe BlackOut in the context of RNNLM training, it can be used to any networks with large softmax output layers.

It is today acknowledged that neural network language models outperform backoff language models in applications like speech recognition or statistical machine translation. However, training these models on large amounts of data can take several days. We present efficient techniques to adapt a neural network language model to new data. Instead of training a completely new model or relying on mixture approaches, we propose two new methods: continued training on resampled data or insertion of adaptation layers. We present experimental results in an CAT environment where the post-edits of professional translators are used to improve an SMT system. Both methods are very fast and achieve significant improvements without overfitting the small adaptation data.

Authorship attribution refers to the task of automatically determining the author based on a given sample of text. It is a problem with a long history and has a wide range of application. Building author profiles using language models is one of the most successful methods to automate this task. New language modeling methods based on neural networks alleviate the curse of dimensionality and usually outperform conventional N-gram methods. However, there have not been much research applying them to authorship attribution. In this paper, we present a novel setup of a Neural Network Language Model (NNLM) and apply it to a database of text samples from different authors. We investigate how the NNLM performs on a task with moderate author set size and relatively limited training and test data, and how the topics of the text samples affect the accuracy. NNLM achieves nearly 2.5% reduction in perplexity, a measurement of fitness of a trained language model to the test data. Given 5 random test sentences, it also increases the author classification accuracy by 3.43% on average, compared with the N-gram methods using SRILM tools. An open source implementation of our methodology is freely available at https://github.com/zge/authorship-attribution/.

Origin and triggers of power quality (PQ) events must be identified in prior, in order to take preventive steps to enhance power quality. However it is important to identify, localize and classify the PQ events to determine the causes and origins of PQ disturbances. In this paper a novel algorithm is presented to classify voltage variations into six different PQ events considering the space phasor model (SPM) diagrams, dual tree complex wavelet transforms (DTCWT) sub bands and the convolution neural network (CNN) model. The input voltage data is converted into SPM data, the SPM data is transformed using 2D DTCWT into low pass and high pass sub bands which are simultaneously processed by the 2D CNN model to perform classification of PQ events. In the proposed method CNN model based on Google Net is trained to perform classification of PQ events with default configuration as in deep neural network designer in MATLAB environment. The proposed algorithm achieve higher accuracy with reduced training time in classification of events than compared with reported PQ event classification methods. 

The optical implementation of weighted interconnections is investigated and basic relationship are derived between the number of neurons, the number of connections and methods for selecting the positions of the neurons to achieve the maximum density of independent connections are presented. The connectivity of a neural network (number of synapses per neuron) is related to the complexity of the problems it can handle. For a network that learns a problem from examples using a local learning rule, it is proved that the entropy of the problem becomes a lower bound for the connectivity of the network.

The optical implementation of weighted interconnections is investigated and basic relationship are derived between the number of neurons, the number of connections and methods for selecting the positions of the neurons to achieve the maximum density of independent connections are presented. The connectivity of a neural network (number of synapses per neuron) is related to the complexity of the problems it can handle. For a network that learns a problem from examples using a local learning rule, it is proved that the entropy of the problem becomes a lower bound for the connectivity of the network.

The optical implementation of weighted interconnections is investigated and basic relationship are derived between the number of neurons, the number of connections and methods for selecting the positions of the neurons to achieve the maximum density of independent connections are presented. The connectivity of a neural network (number of synapses per neuron) is related to the complexity of the problems it can handle. For a network that learns a problem from examples using a local learning rule, it is proved that the entropy of the problem becomes a lower bound for the connectivity of the network.

We investigate the task of building open domain, conversational dialogue systems based on large dialogue corpora using generative models. Generative models produce system responses that are autonomously generated word-by-word, opening up the possibility for realistic, flexible interactions. In support of this goal, we extend the recently proposed hierarchical recurrent encoder-decoder neural network to the dialogue domain, and demonstrate that this model is competitive with state-of-the-art neural language models and back-off n-gram models. We investigate the limitations of this and similar approaches, and show how its performance can be improved by bootstrapping the learning from a larger question-answer pair corpus and from pretrained word embeddings.

The categorization ability of fully connected neural network models, with either discrete or continuous Q-state units, is studied in this work in replica symmetric mean-field theory. Hierarchically correlated multi-state patterns in a two level structure of ancestors and descendents (examples) are embedded in the network and the categorization task consists in recognizing the ancestors when the network is trained exclusively with their descendents. Explicit results for the dependence of the equilibrium properties of a Q=3-state model and a $Q=\infty$-state model are obtained in the form of phase diagrams and categorization curves. A strong improvement of the categorization ability is found when the network is trained with examples of low activity. The categorization ability is found to be robust to finite threshold and synaptic noise. The Almeida-Thouless lines that limit the validity of the replica-symmetric results, are also obtained.

Evaluation of biomedical signals is important in the diagnosis of numerous diseases, chiefly in cardiology through the use of electrocardiograms, and to a more limited extent, in neurology through the use of electroencephalograms. While automated techniques exist for both ECC and EEC analysis, it is likely that additional information can be extracted from these signals through the use of new methods. A chaotic method for analysis of signal analysis variability is presented here that identifies the degree of variability in the signal over time. A second focus is to develop higher order decision models that can incorporate these results with other clinical parameters to represent a more comprehensive view of the disease state, using a neural network model.

Recurrent neural network language models (RNNLMs) have become an increasingly popular choice for speech and language processing tasks including automatic speech recognition (ASR). As the generalization patterns of RNNLMs and n-gram LMs are inherently different, RNNLMs are usually combined with n-gram LMs via a fixed weighting based linear interpolation in state-of-the-art ASR systems. However, previous work doesn't fully exploit the difference of modelling power of the RNNLMs and n-gram LMs as n-gram level changes. In order to fully exploit the detailed n-gram level complementary attributes between the two LMs, a back-off based compact representation of n-gram dependent interpolation weights is proposed in this paper. This approach allows weight parameters to be robustly estimated on limited data. Experimental results are reported on the three tasks with varying amounts of training data. Small and consistent improvements in both perplexity and WER were obtained using the proposed interpolation approach over the baseline fixed weighting based linear interpolation.

This thesis evaluates a possible use of artificial neural networks for military manpower and personnel analysis. Two neural network models were constructed to predict the reenlistment behavior of a select group of individuals in the Navy, from a sample of 680 individuals. The data were extracted from the 1985 DoD Survey of Officer and Enlisted Personnel. Explanatory variables were grouped into demographic/personal, military characteristics, perceived probability of civilian employment, educational level, and satisfaction with military life and military benefits. The first neural network model was compared to a more traditional method of statistical modeling (logistic regression analysis) to determine the strengths and weaknesses of the neural network model. Both models used the same set of 17 variables and were tested using a holdout sample of 100 observations. The neural network model was found to be comparable to the logistic regression model as a predictor, but deficient as a policy analysis model. The second neural network model was constructed using the same data set and architecture as the first neural network model, including the original 17 variables, plus an additional II variables that consisted of variables with and without theoretical foundation for predicting reenlistment. The two neural network models were then compared and found to be similar at predicting reenlistment. Both neural network models were considered to be deficient as tools for policy analysts.... Artificial neural networks, Neural networks, Reenlistment behavior.

Contemporary neural network models provide insights into some of the organizational principles that govern biological sensory-motor systems, and offer a level of computational precision that enables sharp comparisons and contrasts to be made between different sensory-motor systems. The capacity of these models to clarify, integrate, and predict behavioral and neural data is predicated upon the coordinated use of theoretical, mathematical, computational and empirical tools in a manner that reveals many more constraints on brain design than empirical tools alone. No single experimental paradigm in the behavioral and brain sciences provides sufficiently many data to uniquely characterize a neural system. Interdisciplinary theoretical and empirical approaches that can coordinate and discover both top-down and bottom-up constraints at multiple levels of behavioral and neural organization provide a much greater level of guidance towards characterizing brain designs. The present chapter takes as its point of departure one important design principle that has been clarified by such an interdisciplinary approach. This is the principle of vector encoding that has been described, for example, in both the control of saccadic eye movements by the superior colliculus and the control of arm movements by the motor cortex.

We propose Diverse Embedding Neural Network (DENN), a novel architecture for language models (LMs). A DENNLM projects the input word history vector onto multiple diverse low-dimensional sub-spaces instead of a single higher-dimensional sub-space as in conventional feed-forward neural network LMs. We encourage these sub-spaces to be diverse during network training through an augmented loss function. Our language modeling experiments on the Penn Treebank data set show the performance benefit of using a DENNLM.

The article is devoted to the problem automatic decision-making in a mode of real time for difficult objects on basis of neural network emulators. The technology of modeling of base processes of automation of decision-making (recognition of conditions, allocation of the homogeneous data, forecasting of dynamics of behavior) for critical infrastructures in different subject domains is offered. At the heart of the approach there is the idea of creation of the additional, independent tool of decision-making in a mode of real time. The models and algorithms realized in the environment of standard emulators format Statistika Neural Network, providing well-founded indicators of reliability of accepted decisions are resulted. 

This article is from Frontiers in Psychology , volume 4 . Abstract Listening to music often leads to physiological responses. Do these physiological responses contain sufficient information to infer emotion induced in the listener? The current study explores this question by attempting to predict judgments of “felt” emotion from physiological responses alone using linear and neural network models. We measured five channels of peripheral physiology from 20 participants—heart rate (HR), respiration, galvanic skin response, and activity in corrugator supercilii and zygomaticus major facial muscles. Using valence and arousal (VA) dimensions, participants rated their felt emotion after listening to each of 12 classical music excerpts. After extracting features from the five channels, we examined their correlation with VA ratings, and then performed multiple linear regression to see if a linear relationship between the physiological responses could account for the ratings. Although linear models predicted a significant amount of variance in arousal ratings, they were unable to do so with valence ratings. We then used a neural network to provide a non-linear account of the ratings. The network was trained on the mean ratings of eight of the 12 excerpts and tested on the remainder. Performance of the neural network confirms that physiological responses alone can be used to predict musically induced emotion. The non-linear model derived from the neural network was more accurate than linear models derived from multiple linear regression, particularly along the valence dimension. A secondary analysis allowed us to quantify the relative contributions of inputs to the non-linear model. The study represents a novel approach to understanding the complex relationship between physiological responses and musically induced emotion.

In this work the Gardner problem of inferring interactions and fields for an Ising neural network from given patterns under a local stability hypothesis is addressed under a dual perspective. By means of duality arguments an integer linear system is defined whose solution space is the dual of the Gardner space and whose solutions represent mutually unstable patterns. We propose and discuss Monte Carlo methods in order to find and remove unstable patterns and uniformly sample the space of interactions thereafter. We illustrate the problem on a set of real data and perform ensemble calculation that shows how the emergence of phase dominated by unstable patterns can be triggered in a non-linear discontinuous way.

Precise recognition of unknown variables for different types of solar cells is important in design, control, quality, cost estimation, and prediction of solar cell performance. Development of a single solar cell to a set of cells (solar panels) is usually based on a single operating point on the current-voltage characteristic curve. In recent years, a new method to predict cell performance and cell screening by modeling the cell is presented using an equivalent electrical circuit in which each variable corresponds to a physical phenomenon in the solar cell. These analytical models can be represented by a five-variable, seven-variable models, and recently nine-variable model. Due to the nonlinearities and inability of traditional methods in identifying the unknown variables of the system, recently intelligent algorithms have attracted considerable attentions in solving engineering problems. Neural network algorithm (NNA) is a metaheuristic optimization algorithm that is inspired by the function of the neural network of human brain. In this article, the optimum parameter identification technique of a silicon commercial solar cell is used for single diode, two diode, and three diode models. The obtained optimization results of this research are compared with other optimizers in the literature and the surrounding discussions are done. The improvement level reported by the NNA in comparison with the best reported results in the literature for one, two, and three diode models are 0.44, 0.085, and 17.97 percent, respectively. The obtained results of the proposed NNA method have the highest accuracy among the other optimizers in the literature.

Since university enrollment forecasting is very important, many different methods and models have been proposed by researchers. Two new methods for enrollment forecasting are introduced: (1) the fuzzy time series model; and (2) the artificial neural networks model. Fuzzy time series has been proposed to deal with forecasting problems within a fuzzy environment. In this model, the uncertainty encountered in the forecasting process is taken as being produced by our incomplete understanding of nature. As a result, it is different from any stochastic methods. The major problem with this method is that the forecasted values depend to some degree on our interpretations of the outputs of the forecasting model, which makes the process quite subjective. Artificial neural networks represent an advanced technology applied in engineering. When applied in forecasting, uncertainty is ignored. Consequently, the model is a deterministic one. In spite of this, because of the ability to generalize once trained, the network model has robustness. Examples show how these methods are applied. Six figures and two tables illustrate these applications. (SLD)

Mobile healthcare application (mHealth app) assists the frontline health worker in providing necessary health services to the patient. Unfortunately, existing mHealth apps continue to have accuracy issues and limited number of disease detection systems. Thus, an intelligent disease diagnostics system may help medical staff as well as people in poor communities in rural areas. This study proposes a scheme for simultaneously selecting the best neural network models for intelligent disease detection systems on mobile devices. To find the best models for a given dataset, the proposed scheme employs neural network models capable of evolving altered neural network architectures. Eight neural network models are developed simultaneously and then implemented on the android studio platform. Mobile health applications use pre-trained neural network models to provide users with disease prediction results. The performance of the mobile application is measured against the existing available datasets. The trained neural network engines perform admirably, detecting 7 out of 8 diseases with high accuracy ranging from 86% to 100% and a low detection accuracy of 63%. The detection times vary from 0.01 to 0.057 seconds. The developed mHealth app may be used by health workers and patients to improve resource-poor community health services and patients' healthcare quality.

Fine-grained classification is challenging because categories can only be discriminated by subtle and local differences. Variances in the pose, scale or rotation usually make the problem more difficult. Most fine-grained classification systems follow the pipeline of finding foreground object or object parts (where) to extract discriminative features (what). In this paper, we propose to apply visual attention to fine-grained classification task using deep neural network. Our pipeline integrates three types of attention: the bottom-up attention that propose candidate patches, the object-level top-down attention that selects relevant patches to a certain object, and the part-level top-down attention that localizes discriminative parts. We combine these attentions to train domain-specific deep nets, then use it to improve both the what and where aspects. Importantly, we avoid using expensive annotations like bounding box or part information from end-to-end. The weak supervision constraint makes our work easier to generalize. We have verified the effectiveness of the method on the subsets of ILSVRC2012 dataset and CUB200_2011 dataset. Our pipeline delivered significant improvements and achieved the best accuracy under the weakest supervision condition. The performance is competitive against other methods that rely on additional annotations.

In recent years, neural machine translation (NMT) has garnered significant attention due to its superior performance compared to traditional statistical ma chine translation. However, NMT’s effectiveness can be limited when trans lating between languages with dissimilar structures, such as English and Arabic. To address this challenge, recent advances in natural language processing (NLP) have introduced unsupervised pre-training of large neural models, showing promise for enhancing various NLP tasks. This paper proposes a solution that leverages unsupervised pre-training of large neural models to enhance Arabic machine translation (MT). Specifically, we utilize pre-trained checkpoints from publicly available Arabic NLP models, like Arabic bidirectional encoder representations from transformers (AraBERT) and Arabic generative pre-trained transformer (AraGPT), to initialize and warm-start the encoder and decoder of our transformer-based sequence-to-sequence model. This approach enables us to incorporate Arabic-specific linguistic knowledge, such as word morphology and context, into the translation process. Through a comprehensive empirical study, we rigorously evaluated our models against commonly used approaches in Arabic MT. Our results demonstrate that our pre-trained models achieve new state-of-the-art performance in Arabic MT. These findings underscore the effectiveness of pre-trained checkpoints in improving Arabic MT, with potential real-world applications.

Neural Network Algorithms (NNAs), modeled after the workings of biological neurons, are increasingly utilized in areas like data mining and robotics to address complex challenges in artificial intelligence (AI). This research will undertake a systematic review based on advanced neural networks and hybrid models for wind power forecasting. Using the Scopus database, a methodical search, acquisition, and filtering procedure was utilized to find pertinent publication documents; VOSviewer software was utilized to analyze trends. The emphasis on improving prediction accuracy and stability in wind power forecasting through the application of cutting-edge machine learning techniques and hybrid models is a prominent feature that unites the literature. Furthermore, attention is being paid to resolving issues pertaining to the production of wind energy, such as wind power fluctuation management, grid integration problems, wind speed prediction, and turbine health monitoring. A rising trend involves multi-dimensional, multi-step forecasting and incorporating factors like weather data and spatial-temporal features to enhance reliability. This paper contributes by exploring the integration of optimization techniques with neural networks, investigating hybrid models to improve wind power predictions, assessing LSTM-based approaches in forecasting, and suggesting directions for future research.

Considering the crucial importance of nonlinear interaction Snl for the development of third generation wave models the long-term goal of this work is to improve accuracy of calculating nonlinear interaction Snl in wind wave models, and hence improving wave prediction in general.

This research project is aimed at developing silicon based implementations of neural network models. The main advantages of our approach are its use of standard, present day technology and its highly memory due to the use of optics. Two different embodiments of the electronic part of the neural processor have been realized. A phototransistor based network using standard complementary metal oxide semiconductors technology has been built and tested. The CCD version of the optoelectronic architecture has been fabricated and tested, proving the viability of this architecture. All electronic loading has been explored and offers possibilities of rugged, compact systems. (KR)

The grant supported research efforts in two main research questions: (1) How do we segment and estimate the slant/tilt of natural scenes with highly irregular textures, using both biological and non-biological computing architectures; and (2) How do we design -machine perception and action systems and learning mechanisms that can improve their performance through repeated feedback and interaction with their environment? Section 2 describes the research and publications relating to the slant/tilt and segmentation problem. Section 3 describes research and publications using developmental psychological and machine learning frameworks for learning visuomotor tasks. Finally, Section 4 summarizes the two Ph.D dissertations in the Department of Computer and Information Science made possible with AFOSR support.

In this paper, turbulent flow in an asymmetric two-dimensional diffuser is investigated. In many applications, it is important to know whether the boundary layer separates from the surface or inside a particular object, it is also important to know exactly where the flow separation occurs. Combining turbulence data with artificial intelligence is currently an active research topic for studying turbulence. This research makes it possible to replace traditional turbulent models with artificial neural networks (ANN). In this study, to predict flow separation in an asymmetric two-dimensional diffuser, three turbulence models, standard k-, standard k-and SST k-, and intelligent neural network model with reverse pressure gradient were investigated. Fluent software was used to solve the Navier-Stokes-Reynolds equations. Three types of networking are designed and at the end, the second type is used to analyze the flow. 21, 29, 39 and 49 cm distances from the edge of the diffuser were analyzed and compared with experimental data. x and y/H are considered as the input point and U/U0 is the velocity at that point as the output of the neural network model. RMSE, MBE, t-test statistical indices have been calculated and reported for the desired points, The ANN had a better prediction of separation than the other three standard models, and the standard k- had a lower prediction than the other models. This research shows the perspective of chaotic modeling with machine learning methods, especially neural networks.

In this paper, we propose the new fixed-size ordinally-forgetting encoding (FOFE) method, which can almost uniquely encode any variable-length sequence of words into a fixed-size representation. FOFE can model the word order in a sequence using a simple ordinally-forgetting mechanism according to the positions of words. In this work, we have applied FOFE to feedforward neural network language models (FNN-LMs). Experimental results have shown that without using any recurrent feedbacks, FOFE based FNN-LMs can significantly outperform not only the standard fixed-input FNN-LMs but also the popular RNN-LMs.

Organizations of computing elements that follow the principles of physiological neurons, called neural network models, have been shown to have the capability of learning to recognize patterns and to retrieve complete patterns from partial representations. The implementation of neural network models as VLSI or USLI chips within a few years is certain. This report reviews a number of published papers on neural network models and their capabilities. Then, an outline of a speech recognition system that uses neural network modules for learning and recognition is proposed. It is based on the layered structure of existing speech recognition systems, and uses forced learning (feedback) for conditioning the neural modules at the various levels. (Author)

The Sparsely Distributed Memory (SDM) model (Kanerva, 1984) is compared to Hopfield-type neural-network models. A mathematical framework for comparing the two is developed, and the capacity of each model is investigated. The capacity of the SDM can be increased independently of the dimension of the stored vectors, whereas the Hopfield capacity is limited to a fraction of this dimension. However, the total number of stored bits per matrix element is the same in the two models, as well as for extended models with higher order interactions. The models are also compared in their ability to store sequences of patterns. The SDM is extended to include time delays so that contextual information can be used to cover sequences. Finally, it is shown how a generalization of the SDM allows storage of correlated input pattern vectors.

The phenomenon of replica symmetry breaking is investigated for the retrieval phases of Hopfield-type network models. The basic calculation is done for the generalized version of the standard model introduced by Horner [1] and by Perez-Vicente and Amit [2] which can exhibit low mean levels of neural activity. For a mean activity $\bar a =1/2$ the Hopfield model is recovered. In this case, surprisingly enough, we cannot confirm the well known one step replica symmetry breaking (1RSB) result for the storage capacity which was presented by Crisanti, Amit and Gutfreund [3] ($\alpha_c^{\hbox{\mf 1RSB}}\simeq 0.144$). Rather, we find that 1RSB- and 2RSB-Ans\"atze yield only slightly increased capacities as compared to the replica symmetric value ($\alpha_c^{\hbox{\mf 1RSB}}\simeq 0.138\,186$ and $\alpha_c^{\hbox{\mf 2RSB}}\simeq 0.138\,187 $ compared to $\alpha_c^{\hbox{\mf RS}}\simeq 0.137\,905$), significantly smaller also than the value $\alpha_c^{\hbox{\mf sim}} = 0.145\pm 0.009$ reported from simulation studies. These values still lie within the recently discovered reentrant phase [4]. We conjecture that in the infinite Parisi-scheme the reentrant behaviour disappears as is the case in the SK-spin-glass model (Parisi--Toulouse-hypothesis). The same qualitative results are obtained in the low activity range.

Recent studies have shown that deep neural networks (DNN) are vulnerable to adversarial samples: maliciously-perturbed samples crafted to yield incorrect model outputs. Such attacks can severely undermine DNN systems, particularly in security-sensitive settings. It was observed that an adversary could easily generate adversarial samples by making a small perturbation on irrelevant feature dimensions that are unnecessary for the current classification task. To overcome this problem, we introduce a defensive mechanism called DeepCloak. By identifying and removing unnecessary features in a DNN model, DeepCloak limits the capacity an attacker can use generating adversarial samples and therefore increase the robustness against such inputs. Comparing with other defensive approaches, DeepCloak is easy to implement and computationally efficient. Experimental results show that DeepCloak can increase the performance of state-of-the-art DNN models against adversarial samples.

Cellular neural circuit and networks consisting of interconnected neurons and glia are ulti- mately responsible for the information processing associated with information processing in the brain. While there are major efforts aimed at mapping the structural and (electro)physiological connectivity of brain networks, such as the White House BRAIN Initiative aimed at the devel- opment of neurotechnologies capable of high density neural recordings, theoretical and compu- tational methods for analyzing and making sense of all this data seem to be further behind. Here, we propose and provide a summary of an approach for calculating effective connectivity from experimental observations of neuronal network activity. The proposed method operates on network-level data, makes use of all relevant prior knowledge, such as dynamical models of individual cells in the network and the physical structural connectivity of the network, and is broadly applicable to large classes of biological and non-biological networks.

Inferring implicit discourse relations in natural language text is the most difficult subtask in discourse parsing. Surface features achieve good performance, but they are not readily applicable to other languages without semantic lexicons. Previous neural models require parses, surface features, or a small label set to work well. Here, we propose neural network models that are based on feedforward and long-short term memory architecture without any surface features. To our surprise, our best configured feedforward architecture outperforms LSTM-based model in most cases despite thorough tuning. Under various fine-grained label sets and a cross-linguistic setting, our feedforward models perform consistently better or at least just as well as systems that require hand-crafted surface features. Our models present the first neural Chinese discourse parser in the style of Chinese Discourse Treebank, showing that our results hold cross-linguistically.

This article is from Frontiers in Computational Neuroscience , volume 7 . Abstract Human motor adaptation to novel environments is often modeled by a basis function network that transforms desired movement properties into estimated forces. This network employs a layer of nodes that have fixed broad tunings that generalize across the input domain. Learning is achieved by updating the weights of these nodes in response to training experience. This conventional model is unable to account for rapid flexibility observed in human spatial generalization during motor adaptation. However, added plasticity in the widths of the basis function tunings can achieve this flexibility, and several neurophysiological experiments have revealed flexibility in tunings of sensorimotor neurons. We found a model, Locally Weighted Projection Regression (LWPR), which uniquely possesses the structure of a basis function network in which both the weights and tuning widths of the nodes are updated incrementally during adaptation. We presented this LWPR model with training functions of different spatial complexities and monitored incremental updates to receptive field widths. An inverse pattern of dependence of receptive field adaptation on experienced error became evident, underlying both a relationship between generalization and complexity, and a unique behavior in which generalization always narrows after a sudden switch in environmental complexity. These results implicate a model that is flexible in both basis function widths and weights, like LWPR, as a viable alternative model for human motor adaptation that can account for previously observed plasticity in spatial generalization. This theory can be tested by using the behaviors observed in our experiments as novel hypotheses in human studies.

Verification testing is an important aspect of the design process for mechanical mechanisms, and full-scale, full-length life testing is typically used to qualify any new component for use in space. However, as the required life specification is increased, full-length life tests become more costly and lengthen the development time. At the NASA Lewis Research Center, we theorized that neural network systems may be able to model the operation of a mechanical device. If so, the resulting neural network models could simulate long-term mechanical testing with data from a short-term test. This combination of computer modeling and short-term mechanical testing could then be used to verify the reliability of mechanical systems, thereby eliminating the costs associated with long-term testing. Neural network models could also enable designers to predict the performance of mechanisms at the conceptual design stage by entering the critical parameters as input and running the model to predict performance. The purpose of this study was to assess the potential of using neural networks to predict the performance and life of mechanical systems. To do this, we generated a neural network system to model wear obtained from three accelerated testing devices: 1) A pin-on-disk tribometer; 2) A line-contact rub-shoe tribometer; 3) A four-ball tribometer.

Green technology building is not newly introduced to the world nor Malaysia, but it is rarely practiced globally and now it has promoted noteworthy due to destructions caused by human hands towards the nature. Now people started to realize that the world is polluted by many hazardous substances. Therefore, Help University came up with the effort of preserving the nature through a new Green Technology campus, which has been fully operated since year 2017. In this research, neural network forecasting models on energy-efficient data of Help University, Subang 2 green technology campus at Subang Bistari, Selangor has been done with respect to value-for-money (VFM) attribute. Previously there were no similar research done on energy-efficient data of Help University, Subang 2 campus. The significant factors with respect to energy or electricity saved (MW/hr) in the year 2017 variable were studied as recorded by Building Automation and Control System (BAS) of Help University Subang 2 campus. Using multiple linear regression (stepwise method), the significant predictor towards energy saved (MW/hr) was Building Energy Index (BEI) (kWh/m2/year) based p-value<α=0.05. A mathematical model was developed. Moreover, the proposed neural network forecasting model using Killer Whale-Backpropagation Algorithm (KWBP) were found to better than existing conventional techniques to forecast BEI data. This research is expected to specifically assist maintenance department of Help University, Subang 2 campus towards load forecasting for power saving planning in years to come.

In the U.S., there are currently 13.7 million cancer survivors (38). Many cancer survivors experience problems with post-treatment mental and physical functioning.Although research has identified important contributing factors regarding these problems, traditional predictive statistical modeling accounts for less than half the variance in mental and physical function (16; 17; 113). The relationship among these factors may be better accounted for by a non-linear modeling approach. The goal of this doctoral study was to determine whether a non-linear, adaptive predictive model demonstrated better model fit, showed greater predictive accuracy, and accounted for a greater contribution of independent variables over a traditional statistical model with regard to mental and physical functioning in post-treatment breast cancer survivors. Using demographic, medical, and clinical variables, linear regression was compared to neural network modeling in predicting mental functioning and physical functioning in a sample of post-treatment breast cancer survivors. Contrary to the a priori hypotheses, the neural network model did not outperform the linear regression model in predicting mental and physical functioning of post-treatment breast cancer survivors. However, both linear regression and neural network modeling identified modifiable variables (clinical domains of the Cancer Survivor Profile) as important predictors of post-treatment mental and physical functioning, with the neural network confirming the findings of the linear regression models. The neural network model also added to the results of the linear regression by identifying additional important variables (age, time since diagnosis) that may have a non-linear relationship with mental and physical functioning. These findings may promote a better understanding of post-treatment health status and promote targeted clinical interventions.

Presently, the mesh embedment in masonry is becoming a trendy research topic. In this paper, the mesh embedded masonry prism was cast and tested. The experimental data were used for the analytical modelling. Compressive strength (CS) test was conducted for forty five masonry prism specimens with and without poultry netting mesh (PNM) embedment in the bed joints. The small mesh embedment in the masonry prism provides the better strength improvement as well as the endurance. The size of masonry prism was 225×105×176 mm. Uniformity was maintained in all prisms as per the guidelines given in ASTM C1314. Compressive strength experimental results are compared with a new proposed regression equation. The equation needs nine input parameters and two adjustment coefficients. The masonry mortar strength and mesh embedment are considered as input parameter. The experimental results were predicted by proposed Artificial Neural Network model. The validated results were gives better and more accuracy compared to the statistical and MLRPM models.

This study develops prediction models using Artificial Neural Network (ANN) to describe the long term performance of reinforcements in concrete slabs containing pure Magnesium anodes and subjected to chloride ingress. Ten reinforced concrete slabs of dimensions 1000 mm x 1000 mm x 100 mm were cast. Five slabs were cast with 3.5% NaCl by weight of cement, and five more were cast without NaCl. The distance of the point under consideration from the anode in the x and y axes, temperature, relative humidity (RH), and age of concrete in days were considered input parameters, while the half-cell potential (HCP) values with reference to the Standard Calomel Electrode (SCE) were considered output. Experimental values consisting of 80 HCP values per slab per day were collected for 270 days and were averaged for both cases to generate the prediction model. Various learning heuristics used in supervised learning in feedforward ANN was used (viz. RB, OSS, SCG, GDA, CGP, CGF, GDX, and LM). A two-layer feed-forward network with 10 sigmoid hidden neurons and trained linear output neurons was employed in this work. The network architecture [5-10-1] and 10 neurons in the hidden layer were used for all the prediction models. Out of all the training algorithms used, the overall performance was best with LM in all stages of modelling (>96%). To conclude, the prediction of HCP values through the neural network models based on the available experimental data set was excellent.

Since the emergence of Deep Neural Networks (DNNs) as a prominent technique in the field of computer vision, the ImageNet classification challenge has played a major role in advancing the state-of-the-art. While accuracy figures have steadily increased, the resource utilisation of winning models has not been properly taken into account. In this work, we present a comprehensive analysis of important metrics in practical applications: accuracy, memory footprint, parameters, operations count, inference time and power consumption. Key findings are: (1) power consumption is independent of batch size and architecture; (2) accuracy and inference time are in a hyperbolic relationship; (3) energy constraint is an upper bound on the maximum achievable accuracy and model complexity; (4) the number of operations is a reliable estimate of the inference time. We believe our analysis provides a compelling set of information that helps design and engineer efficient DNNs.

Only a few parameters that affect the machining process in ECM are controllable. It is clearly established from reported works that results reported in literature cannot be extrapolated. So for any new material - electrolyte combination and machining conditions experiments need to be conducted to predict the effects of process parameters on machined geometry. For effective exploitation of ECM for machining SG Iron it is essential to develop models for predicting the nature of surface that will be generated. Two widely used methods for correlating controllable process parameters and surface roughness parameters are Design of Experiments and Artificial Neural Networks (ANN). The back–propagation algorithm is used widely as a learning algorithm in feed-forward multilayer neural network. A large data set is usually necessary to train the ANNs. However, generating a large number of data for training ANN is not only time consuming but very expensive. Hence to reduce the cost of developing NN models it is decided to train the network with only those thirteen data that are used for developing Box Behnken design based models. The three configurations of the ANN considered are 3-10-10-1, 3-20-20-20-1 and3-39-39-39-1. Because of the small data set the deviations noted between predictions obtained using Box-Behnken models and ANNs are high. Denoising is done in two stages using Multiscale Principal Component Analysis (PCA) and dropping the high frequency coefficients by filtering with Daubechies wavelet. The first stage denoising carried out using PCA has resulted in significant reduction in deviation (error). After second level denoising the errors have reduces further. Of the three network configurations studied lowest configuration gave the best results in majority of the cases and where as for the highest configuration, the denoising effect is small. Not only the error is reduced but the number of points with large errors has reduced substantially. The differences between the outputs from the neural network model and Box Behnken models are well within the confidence intervals calculated from ANOVA.

Transliteration is a key component of machine translation systems and software internationalization. This paper demonstrates that neural sequence-to-sequence models obtain state of the art or close to state of the art results on existing datasets. In an effort to make machine transliteration accessible, we open source a new Arabic to English transliteration dataset and our trained models.

Information retrieval in a 'connectionist' or neural network is viewed as computing the most probable value of the information to be retrieved with respect to a probability density function, P. With a minimal number of assumptions, the 'energy' function that a neural network minimizes during information retrieval is shown to uniquely specify P. Inspection of the form of P indicates the class of probabilistic environments that can be learned. Learning algorithms can be analyzed and designed by using maximum likelihood estimation techniques to estimate the parameters of P. The large class of nonlinear auto-associative networks analyzed by Cohen and Grossberg (1983), nonlinear associative multi-layer back-propagation networks (Rumelhart, Hinton, & Williams, 1986), and certain classes of nonlinear multi-stage networks are analyzed within the proposed computational framework. Keywords: Artificial intelligence, Connectionism, Non-linear associator.

We have recently shown that deep Long Short-Term Memory (LSTM) recurrent neural networks (RNNs) outperform feed forward deep neural networks (DNNs) as acoustic models for speech recognition. More recently, we have shown that the performance of sequence trained context dependent (CD) hidden Markov model (HMM) acoustic models using such LSTM RNNs can be equaled by sequence trained phone models initialized with connectionist temporal classification (CTC). In this paper, we present techniques that further improve performance of LSTM RNN acoustic models for large vocabulary speech recognition. We show that frame stacking and reduced frame rate lead to more accurate models and faster decoding. CD phone modeling leads to further improvements. We also present initial results for LSTM RNN models outputting words directly.

Distant speech recognition is a challenge, particularly due to the corruption of speech signals by reverberation caused by large distances between the speaker and microphone. In order to cope with a wide range of reverberations in real-world situations, we present novel approaches for acoustic modeling including an ensemble of deep neural networks (DNNs) and an ensemble of jointly trained DNNs. First, multiple DNNs are established, each of which corresponds to a different reverberation time 60 (RT60) in a setup step. Also, each model in the ensemble of DNN acoustic models is further jointly trained, including both feature mapping and acoustic modeling, where the feature mapping is designed for the dereverberation as a front-end. In a testing phase, the two most likely DNNs are chosen from the DNN ensemble using maximum a posteriori (MAP) probabilities, computed in an online fashion by using maximum likelihood (ML)-based blind RT60 estimation and then the posterior probability outputs from two DNNs are combined using the ML-based weights as a simple average. Extensive experiments demonstrate that the proposed approach leads to substantial improvements in speech recognition accuracy over the conventional DNN baseline systems under diverse reverberant conditions.

This paper presents a novel latent variable recurrent neural network architecture for jointly modeling sequences of words and (possibly latent) discourse relations between adjacent sentences. A recurrent neural network generates individual words, thus reaping the benefits of discriminatively-trained vector representations. The discourse relations are represented with a latent variable, which can be predicted or marginalized, depending on the task. The resulting model can therefore employ a training objective that includes not only discourse relation classification, but also word prediction. As a result, it outperforms state-of-the-art alternatives for two tasks: implicit discourse relation classification in the Penn Discourse Treebank, and dialog act classification in the Switchboard corpus. Furthermore, by marginalizing over latent discourse relations at test time, we obtain a discourse informed language model, which improves over a strong LSTM baseline.

The analysis of large-scale simulations can pose a large computational burden, often necessitating the use of high performance computers. Moreover, these models can be analytically intractable because of complex, internal logical relationships. To begin to overcome these types of obstacles, a method known as meta-modeling has been developed to construct mathematical functions that act as analytic surrogates to large scale simulations. This research examines the introduction of second-order interactions for two types of asymptotically-standardized linear control variates to least squares regression and radial basis neural network meta-models for a queuing simulation. Extensions are made to the statistical framework for variance reduction of direct estimation of single response, single population simulations and the framework for meta-models of single response, multiple population simulations. As a result, the new extensions are shown to significantly outperform existing frameworks and also provide the means to interpret and better understand the system dynamics that manifest when a system exhibits an exceptional amount of variance.

Open-ended learning environments (OELEs) allow students to freely interact with the content and to discover important principles and concepts of the learning domain on their own. However, only some students possess the necessary skills for efficient and effective exploration. Guidance in the form of targeted interventions or feedback therefore has the potential to improve educational outcomes. A promising approach for adaptation in OELEs is the design of interventions based on the detection of characteristic learning behaviors through offline clustering, followed by a real-time classification of new students. In this paper, we explore the possibility of using recurrent neural network (RNN) models for this online classification task. We extensively evaluate the predictive performance of different variants of RNNs, namely long-short term memory models and gated recurrent units, and different architectures on a data set collected from an exploration-based educational game. We also compare the prediction accuracy of the different RNN models to the performance of traditional classifiers on the same data set. Our results demonstrate that RNNs perform similar or better than traditional methods regarding early classification and therefore constitute a promising alternative for the online classification of new students. [For the full proceedings, see ED599096.]

1Canterville Ghost

By Oscar Wilde

The American Minister and his family have bought the English stately home Canterville Chase, complete with the ghost of Sir Simon de Canterville - blood-stains, clanking chains and all. But these modern Americans will have no truck with ghostly goings-on, and set out to beat the spectre at his own game. (Summary by David Barnes)

2Importance of Being Earnest

By Oscar Wilde

The Importance of Being Earnest is a classic comedy of manners in which two flippant young men, in order to impress their respected beloveds, pretend that their names are “Ernest,” which both young ladies believe confers magical qualities on the possessor. It was first performed for the public on February 14, 1895 at the St. James’ Theatre in London, and is regarded by many critics and scholars as being the wittiest play in the English language. (Summary from Wikipedia.org)<br /><br /><b>Characters:</b><br /><em>John Worthing, J.P.</em> - Read by <a href="http://www.johngon.com" rel="nofollow">John Gonzalez</a><br /><em>Algernon Moncrieff</em> - Read by <a href="http://www.gordmackenzie.com/" rel="nofollow">Gord Mackenzie</a><br /><em>Rev. Canon Chasuble, D.D.</em> - Read by <a href="http://tis.goringe.net/" rel="nofollow">Chris Goringe</a><br /><em>Merriman, Butler</em> - Read by Peter Yearsley<br /><em>Lane, Manservant</em> - Read by <a href="http://dosemagazine.blogsome.com/" rel="nofollow">Hugh McGuire</a><br /><em>Lady Bracknell</em> - Read by <a href="http://www.mediatinker.com/" rel="nofollow">Kristen McQuillin</a><br /><em>Hon. Gwendolen Fairfax</em> - Read by Deana<br /><em>Cecily Cardew</em> - Read by <a href="http://kayray.org/" rel="nofollow">Kara Shallenberg</a><br /><em>Miss Prism, Governess</em> - Read by Sureka<br /><em>Narrator</em> - Read by <a href="http://betsie.info" rel="nofollow">Betsie Bush</a><br /><em>Project Editor</em> - <a href="http://www.writingshow.com/" rel="nofollow">Paula Berinstein</a>

3Picture of Dorian Gray

By Oscar Wilde

Dorian Gray, a young man of wealth and stature in late 1800's London, meets Lord Henry Wotton while posing for a portrait by his friend Basil Hallward. Once the painting is complete, Dorian realizes that it will always be young and attractive, while he will be forced to age and wither with the years. Carelessly, he wishes the opposite were true. What happens is a treatise on morals, self-indulgence and how crucial personal responsibility is towards one's self. (Summary by John Gonzalez)

4Happy Prince and Other Tales

By Oscar Wilde

Collection of children’s stories written in 1888, dealing primarily with love and selfishness. These stories are generally sad, with a moralistic message. The collection includes: The Happy Prince, The Nightingale and the Rose, The Selfish Giant, The Devoted Friend, and The Remarkable Rocket. (Summary written by Joy Chan)

5Reviews

By Oscar Wilde

Wilde’s literary reputation has survived so much that I think it proof against any exhumation of articles which he or his admirers would have preferred to forget. As a matter of fact, I believe this volume will prove of unusual interest; some of the reviews are curiously prophetic; some are, of course, biassed by prejudice hostile or friendly; others are conceived in the author’s wittiest and happiest vein; only a few are colourless. And if, according to Lord Beaconsfield, the verdict of a continental nation may be regarded as that of posterity, Wilde is a much greater force in our literature than even friendly contemporaries ever supposed he would become. <p>It should be remembered, however, that at the time when most of these reviews were written Wilde had published scarcely any of the works by which his name has become famous in Europe, though the protagonist of the æsthetic movement was a well-known figure in Paris and London. (Summary from Introduction by Robert Ross)</p>

6Ballad of Reading Gaol

By Oscar Wilde

Wilde’s meditation on capital punishment, the Ballad of Reading Gaol comes after he was convicted and imprisoned under charges of gross indecency. The charges stemmed from his affair with Lord Alfred Douglas, the son of the Marquis of Queensberry. It relates the story of an execution of a man who murdered his wife which Wilde witnessed during his internment. Published in 1898, it was Wilde’s last published poem as he would die in 1900 from cerebral meningitis, caused by syphilis.(Summary by John Gonzalez)

7Ballad of Reading Gaol, (version 2)

By Oscar Wilde

In 1895, Oscar Wilde was sentenced to 2 years of hard labor for acts of ‘gross indecency’. During his time at Reading Gaol, he witnessed a rare hanging, and in the three years between his release and his untimely death in 1900, was inspired to write the following poem, a meditation on the death penalty and the importance of forgiveness, even for (and especially for) something as heinous as murdering one’s spouse; for even the murderer, Wilde argues, is human and suffers more so for being the cause of his own pain, for ‘having killed the thing he loved’; for everyone is the cause of someone else’s suffering and suffers at the hands of another. It is this that Jesus Christ could see; he could continue to see the beauty of our humanity, despite all that we may do to each other, and encouraged us to love each other just the same<br /><br /> “The Ballad of Reading Gaol” was published in 1898 and would gain Wilde greater recognition as a poet (in addition to being a great playwright); although his only other volume of poetry, one of his earliest works that he’d published, was also well-received. Sadly, ‘The Ballad’ would be his last.<br />(Summary by Linda Leu).

8Ideal Husband

By Oscar Wilde

<i>An Ideal Husband</i> is an 1895 comedy by Oscar Wilde which revolves around blackmail and political corruption, and touches on the themes of public and private honor. The action is set in London, in "the present", and takes place within a single day. "Sooner or later," Wilde notes, "we shall all have to pay for what we do." But he adds that, "No one should be entirely judged by their past." (Summary from Wikipedia)<br /><br /> <b>Characters:</b><br /><em>Narrator</em> &#8211; Read by <a href="http://alexfoster.me.uk/" rel="nofollow">Alex Foster</a><br /><em>The Earl Of Caversham, K.G.</em> &#8211; Read by Linton<br /><em>Viscount Goring, his son</em> &#8211; Read by John Gonzalez<br /><em>Sir Robert Chiltern, Bart., Under-Secretary for Foreign Affairs</em> &#8211; Read by <a href="http://www.earthcallingdavid.com/" rel="nofollow">David Barnes</a><br /><em>Vicomte De Nanjac, Attache at the French Embassy in London</em> &#8211; Read by Juho Fröjd<br /><em>Mr. Montford</em> &#8211; Read by <a href="http://matthew-walton.co.uk/" rel="nofollow">Matthew Walton</a><br /><em> Mason, Butler to Sir Robert Chiltern </em> &#8211; Read by <a href="http://dosemagazine.blogsome.com/" rel="nofollow">Hugh McGuire</a> <br /><em>Phipps, Lord Goring&#8217;s servant</em> &#8211; Read by Juho Fröjd<br /><em> James, footman</em> &#8211; Read by Read by <a href="http://tis.goringe.net/" rel="nofollow">Chris Goringe</a><br /><em>Harold, footman</em> &#8211; Read by Zachary Brewster-Geisz<br /><em>Lady Chiltern</em> &#8211; Read by <a href="http://librivox.org/newcatalog/people_public.php?peopleid=28" rel="nofollow">Kristin Hughes</a><br /><em>Lady Markby</em> &#8211; Read by Heather Barnett<br /><em>The Countess Of Basildon</em> &#8211; Read by JemmaBlythe<br /><em>Mrs. Marchmont</em> &#8211; Read by <a href="http://librivox.org/wiki/moin.cgi/ezwa" rel="nofollow"> Ezwa</a><br /><em>Miss Mabel Chiltern, Sir Robert Chiltern&#8217;s sister</em> &#8211; Read by Ophelia Darcy<br /><em>Mrs. Cheveley</em> &#8211; Read by <a href="http://betsie.info" rel="nofollow">Betsie Bush</a><br /><em>Project Editor</em> &#8211; <a href="http://www.writingshow.com/" rel="nofollow">Paula Berinstein</a>

9Lady Windermere's Fan

By Oscar Wilde

<i>Lady Windermere's Fan: A Play About a Good Woman</i> is a four act comedy by Oscar Wilde, published in 1893. As in some of his other comedies, Wilde satirizes the morals of Victorian society, and attitudes between the sexes.<br /><br />The action centres around a fan given to Lady Windermere as a present by her husband, and the ball held that evening to celebrate her 21st birthday. (Summary by Ruth Golding)<br /><br /> <strong>Cast:</strong><br /> Lady Windermere - <a href="http://librivox.org/reader/28">Kristin Hughes</a><br /> Lord Windermere - <a href="http://librivox.org/reader/2195">Joe Earley</a><br /> Mrs. Erlynne - <a href="http://librivox.org/reader/1259">Elizabeth Klett</a><br /> Lord Darlington - <a href="http://librivox.org/reader/827">Simon Taylor</a><br /> The Duchess of Berwick - <a href="http://librivox.org/reader/2607">Ruth Golding</a><br /> Mr. Cecil Graham - <a href="http://librivox.org/reader/2358">Chugosh</a><br /> Lord Augustus Lorton - <a href="http://librivox.org/reader/2538">Chrystian J. Gonzalez</a><br /> Mr. Dumby - <a href="http://librivox.org/reader/1913">Geoff Cowgill</a><br /> Parker - <a href="http://librivox.org/reader/2427">Ed Meade</a><br /> Lady Agatha Carlisle - <a href="http://librivox.org/reader/123">Ezwa</a><br /> Lady Plymdale - <a href="http://librivox.org/reader/84">Christie Nowak</a><br /> Mr. Hopper - <a href="http://librivox.org/reader/1094">tubeyes</a><br /> Lady Jedburgh - <a href="http://librivox.org/reader/1649">Kalynda</a><br /> Rosalie - <a href="http://librivox.org/reader/1328">Duan</a><br /> Lady Stutfield - <a href="http://librivox.org/reader/1975">Sara Korsmo</a><br /> Mrs. Cowper-Cowper - <a href="http://librivox.org/reader/2592">Anna Roberts</a><br /> Narrator - <a href="http://librivox.org/reader/2748">Annoying Twit</a> <br /><br /> Audio edited by Rosalind Wills and Ruth Golding.

10De Profundis

By Oscar Wilde

This short work of Wilde's was written during his two year incarceration for "gross indecency". This work is a letter which sorts out his life, and his love toward Lord Alfred Douglas. Wilde wrote this as a farewell letter to Douglas. (summary by Aaron Elliott)<br /><br /> NOTE: "Transcribed from the 1913 Methuen & Co. edition. Note that later editions of De Profundis contained more material. The most complete editions are still in copyright in the U.S.A."

In 1894, Oscar Wilde (1854 - 1900) published two collections of aphorisms: <em>A Few Maxims For The Instruction Of The Over-Educated</em>, in the Saturday Review newspaper, and <em>Phrases and Philosophies for the Use of the Young</em>, in the Oxford student magazine The Chameleon. By turns witty, intellectual, counter-intuitive and obtruse, the collections came to be seen by many as emblematic of Wilde's style, and countless collections of Wildean aphorisms have since been published. (Summary by Carl Manchester)

LibriVox volunteers bring you 12 different recordings of <em>Serenade</em> by Oscar Wilde. This was the weekly poetry project for the week of April 20th, 2008.

13Importance of Being Earnest (version 2)

By Oscar Wilde

Oscar Fingal O’Flahertie Wills Wilde (1854 – 1900) was an Irish playwright, novelist, poet, and author of short stories. Known for his barbed wit, he was one of the most successful playwrights of late Victorian London, and one of the greatest celebrities of his day. <br /><br />In Wilde’s classic play The Importance of Being Earnest, Jack Worthing and Algernon Moncrieff discover the perils of love, assumed identities, and telling the truth. (Summary by Toby Paradis and Wikipedia)<br /><br /> <b>LibriVox cast:</b><br /> John Worthing, J.P.: <a href="https://librivox.org/reader/2700">Hans-Stefan Ducharme</a><br>Algernon Moncrieff: <a href="https://librivox.org/reader/2697">Simon Ferland</a><br>Rev. Canon Chasuble, D.D.: <a href="https://librivox.org/reader/2280">Toby Paradis</a><br>Merriman, Butler: <a href="https://librivox.org/reader/2280">Toby Paradis</a><br>Lane, Manservant: <a href="https://librivox.org/reader/2280">Toby Paradis</a><br>Lady Bracknell: <a href="https://librivox.org/reader/2698">Sarah A. Farnham</a><br>Hon. Gwendolen Fairfax: <a href="https://librivox.org/reader/2701">Nyssa Gatcombe</a><br>Cecily Cardew: <a href="https://librivox.org/reader/2699">Rebecca Bailey</a><br>Miss Prism, Governess: <a href="https://librivox.org/reader/2702">Eileen Nadeau</a><br />Sound Tech - Keagan Rae

14Picture of Dorian Gray (Lippincott)

By Oscar Wilde

The Picture of Dorian Gray is the only published novel written by Oscar Wilde, and first came out as the lead story in Lippincott's Monthly Magazine on June 20, 1890. Wilde later revised this edition, making several alterations, and adding new chapters; the amended version was published by Ward, Lock, and Company in April 1891. The novel tells of a young man named Dorian Gray, the subject of a painting by artist Basil Hallward. Basil is greatly impressed by Dorian's physical beauty and becomes strongly infatuated with him, believing that his beauty is responsible for a new mode in his art. Talking in Basil's garden, Dorian meets Lord Henry Wotton, a friend of Basil's, and becomes enthralled by Lord Henry's world view. Espousing a new kind of hedonism, Lord Henry suggests that the only thing worth pursuing in life is beauty, and the fulfilment of the senses. Realising that one day his beauty will fade, Dorian cries out, wishing that the portrait Basil has painted of him would age rather than himself. Dorian's wish is fulfilled, subsequently plunging him into a series of debauched acts. The portrait serves as a reminder of the effect each act has upon his soul, with each sin being displayed as a disfigurement of his form, or through a sign of aging. The Picture of Dorian Gray is considered one of the last works of classic gothic horror fiction with a strong Faustian theme. It deals with the artistic movement of the decadents, and homosexuality, both of which caused some controversy when the book was first published. However, in modern times, the book has been referred to as "one of the modern classics of Western literature." (Summary by Wikipedia)

15Florentine Tragedy and La Sainte Courtisane

By Robert Baldwin Ross and Oscar Wilde

Two short fragments: an unfinished and a lost play. A Florentine Tragedy, left in a taxi (not a handbag), is Wilde’s most successful attempt at tragedy – intense and domestic, with surprising depth of characterisation. It was adapted into an opera by the Austrian composer Alexander Zemlinsky in 1917. La Sainte Courtisane, or The Woman Covered in Jewels explores one of Wilde’s great idées fixes: the paradox of religious hedonism, pagan piety. Both plays, Wildean to their core, revel in the profound sadness that is the fruit of the conflict between fidelity and forbidden love. Written towards the end of his tragic life, these fragments give us a glimpse of a genius at his best: visceral, passionate, personal, poetic. (Summary by Simon Larois)<br><br> A Florentine Tragedy - cast:<BR> Narrator: <a href="https://catalog.librivox.org/people_public.php?peopleid=3157">TriciaG</a><br>GUIDO BARDI, A Florentine prince: <a href="https://catalog.librivox.org/people_public.php?peopleid=1492">mb</a><br>SIMONE, a merchant: <a href="https://catalog.librivox.org/people_public.php?peopleid=2986">Simon Larois</a><br>BIANNA, his wife: <a href="https://catalog.librivox.org/people_public.php?peopleid=2607">Ruth Golding</a><br> <BR> La Sainte Courtisane - cast:<BR> Narrator: <a href="https://catalog.librivox.org/people_public.php?peopleid=2607">Ruth Golding</a><br>First man: <a href="https://catalog.librivox.org/people_public.php?peopleid=1492">mb</a><br>Myrrhina: <a href="https://catalog.librivox.org/people_public.php?peopleid=2297">Philippa</a><br>Second man: <a href="https://catalog.librivox.org/people_public.php?peopleid=3332">L.French</a><br>Honorius: <a href="https://catalog.librivox.org/people_public.php?peopleid=3098">woggy298</a><br> <BR> Edited by <a href="https://catalog.librivox.org/people_public.php?peopleid=2607">Ruth Golding</a>

16Bildnis des Dorian Gray

By Oscar Wilde

Als der berückend gutaussehende Jüngling Dorian Gray sein Porträt betrachtet, kommt ihm der Gedanke, wie herrlich es sein müsste, wenn das Bild an seiner Stelle altern könnte, während er selbst für immer jung und schön bliebe. Als sich dieser fantastische Wunsch zu erfüllen beginnt, genießt Dorian seine ewige Jugend in vollen Zügen - während in seiner Seele ein grauenvolles Werk der Zerstörung seinen Lauf nimmt. (Zusammenfassung von Al-Kadi)

17Fisherman and his Soul

By Oscar Wilde

”The Fisherman and his Soul” is a fairy tale first published in November of 1891 in Wilde’s “A House of Pomegranates”. It tells of a fisherman who nets and falls in love with a mermaid. But to be with her he must shed his soul, which goes off to have adventures of its own. Will forbidden love endure? -- This reader has also recorded the three other tales in “A House of Pomegranates” which can be found in the LibriVox Short Story Collections, volumes 37, 38, and 39. (Summary by Gregg Margarite)

The original 1891 version of the play was in French. Three years later an English translation was published. The play tells in one act the Biblical story of Salome, stepdaughter of the tetrarch Herod Antipas, who, to her stepfather's dismay but to the delight of her mother Herodias, requests the head of Iokanaan (John the Baptist) on a silver platter as a reward for dancing the Dance of the Seven Veils. (Summary by wikipedia)<br /><br /> <strong>Cast:</strong><br /> Herod Antipas: <a href="http://librivox.org/reader/2149">Mark Penfold</a><br /> Iokanaan: <a href="http://librivox.org/reader/3699">Bruce Pirie</a><br /> The Young Syrian: <a href="http://librivox.org/reader/4358">Kim Stich</a><br /> Tigellinus: <a href="http://librivox.org/reader/5635">Matthew Reece</a><br /> A Cappadocian: <a href="http://librivox.org/reader/5386">Tim Ferreira</a><br /> A Nubian: <a href="http://librivox.org/reader/5303">D MoNeY</a><br /> First Soldier: <a href="http://librivox.org/reader/5346">Skythrock</a><br /> Second Soldier: <a href="http://librivox.org/reader/5404">Joshua Logan</a><br /> The Page of Herodias: <a href="http://librivox.org/reader/4572">om123</a><br /> Herodias: <a href="http://librivox.org/reader/5285">Sweetlilbirdy</a><br /> Salome: <a href="http://librivox.org/reader/3536">Arielle Lipshaw</a><br /> A Slave: <a href="http://librivox.org/reader/5103">Ana</a><br /> First Jew: <a href="http://librivox.org/reader/5506">Frank Booker</a><br /> Second Jew/Fourth Jew/Sadducee: <a href="http://librivox.org/reader/1259">Elizabeth Klett</a><br /> Third Jew/First Nazarene: <a href="http://librivox.org/reader/26">Denny Sayers</a><br /> Fifth Jew: <a href="http://librivox.org/reader/2911">David Lawrence</a><br /> Second Nazarene: <a href="http://librivox.org/reader/4705">Algy Pug</a><br /> A Pharisee: <a href="http://librivox.org/reader/1492">mb</a><br /> Narrator: <a href="http://librivox.org/reader/3157">Tricia G</a><br /><br /> <strong>Audio edited by:</strong> Arielle Lipshaw

19Picture of Dorian Gray (version 2 dramatic reading)

By Oscar Wilde

The novel tells of a young man named Dorian Gray, the subject of a painting by artist Basil Hallward. Basil is impressed by Dorian's beauty and becomes infatuated with him, believing his beauty is responsible for a new mode in his art. Dorian meets Lord Henry Wotton, a friend of Basil's, and becomes enthralled by Lord Henry's world view. Espousing a new hedonism, Lord Henry suggests the only things worth pursuing in life are beauty and fulfillment of the senses. Realizing that one day his beauty will fade, Dorian expresses his desire to sell his soul to ensure the portrait Basil has painted would age rather than himself. Dorian's wish is fulfilled, plunging him into debauched acts. The portrait serves as a reminder of the effect each act has upon his soul, with each sin displayed as a disfigurement of his form, or through a sign of aging. This reading uses the 20-chapter 1891 version of Wilde's novel. (Summary by Wikipedia)<br /><br/> <strong>Cast:</strong><br/><br/>NARRATOR - <a > Martin Geeson</a><br /> Lord Henry Wotton - <a > David Goldfarb</a><br /> Dorian Gray - <a > Volunteer </a><br /> Basil Hallward - <a > Anthony</a><br /> Sibyl Vane - <a > Miss Avarice</a><br /> James Vane - <a > David Lawrence</a><br /> Duchess of Monmouth - <a > Availle</a><br /> Victor - <a > Martin Geeson</a><br /> Francis - <a > Simon Pride</a><br /> Cab Driver - <a > Simon Pride</a><br /> Parker - <a > Elizabeth Klett</a><br /> Lord Fermor - <a > Anthony</a><br /> Lady Agatha - <a > Sarah</a><br /> Duchess of Harley - <a > Hannah Harris</a><br /> Sir Thomas Burdon - <a > Terence Taylor</a><br /> Mr. Erskine - <a > Frank Booker</a><br /> Mrs. Vandeleur - <a > Mary-Beth Blackburn</a><br /> Lady Henry - <a > Susanna</a><br /> Mrs. Vane - <a > Arielle Lipshaw</a><br /> Mrs. Leaf - <a > Rebeka Harris</a><br /> Mr. Hubbard - <a > Frank Booker</a><br /> Alan Campbell - <a > Ernst Pattynama</a><br /> Lady Narborough - <a > Elizabeth Klett</a><br /> Lady Ruxton - <a > Mary-Beth Blackburn</a><br /> Adrian Singleton - <a > Joseph Lawler</a><br /> Woman - <a > Lucy Perry</a><br /> Sir Geoffrey Clouston - <a > Mark F. Smith</a><br /> Gamekeeper - <a > Martin Geeson</a><br /> Gardener- <a > Joseph Lawler</a><br /> Young Man - <a > Elizabeth Klett</a><br /> Old Gentleman - <a > Mark F. Smith</a><br /> Constable - <a > Joseph Lawler</a><br /> <br />Audio edited by Arielle Lipshaw

20Soul of Man

By Oscar Wilde

“(T)he past is what man should not have been. The present is what man ought not to be. The future is what artists are.”<br/><br/> Published originally as “The Soul of Man Under Socialism,” this is not so much a work of sober political analysis; rather it can be summed up as a rhapsodic manifesto on behalf of the Individual. Socialism having deployed technology to liberate the whole of humanity from soul-destroying labour, the State obligingly withers away to allow the free development of a joyful, anarchic hedonism...<br/><br/> “Is this Utopian? A map of the world that does not include Utopia is not worth even glancing at, for it leaves out the one country at which Humanity is always landing.”<br/><br/> Far from abandoning the epigram in favour of the slogan, Wilde wittily assails several of his favourite targets: the misguided purveyors of philanthropy; life-denying ascetics of various kinds; the army of the half-educated who constitute themselves the enemies of Art - and those venal popular journalists who cater to them...<br/><br/> “Behind the barricade there may be much that is noble and heroic. But what is there behind the leading-article but prejudice, stupidity, cant, and twaddle?” (Introduction by Martin Geeson)

21Happy Prince and Other Tales (version 2)

By Oscar Wilde

The Happy Prince and Other Tales (also sometimes called The Happy Prince and Other Stories) is an 1888 collection of stories for children by Oscar Wilde. It is most famous for The Happy Prince, the short tale of a metal statue who befriends a migratory bird. Together, they bring happiness to others, in life as well as in death.<br /><br /> The stories convey an appreciation for the exotic, the sensual and for masculine beauty. (summary from Wikipedia.)

22De Profundis (version 2)

By Oscar Wilde

This is a letter written from prison in 1897 by Oscar Wilde to Lord Alfred Douglas, in which he recounts how he came to be in prison and charts his spiritual development. (summary by AdamH)</p> NOTE: ."Transcribed from the 1913 Methuen & Co. edition Note that later editions of De Profundis contained more material. The most complete editions are still in copyright in the U.S.A."

23Woman of No Importance

By Oscar Wilde

A Woman of No Importance is a play by Irish playwright Oscar Wilde. The play premièred on 19 April 1893 at London's Haymarket Theatre. It is a testimony of Wilde's wit and his brand of dark comedy. It looks in particular at English upper class society and has been reproduced on stages in Europe and North America since his death in 1900. (Summary by Wikipedia) <br><br><strong>Cast:</strong><br>Lord Illingworth: <a href="https://catalog.librivox.org/people_public.php?peopleid=6037">Peter Bishop</a><br>Sir John Pontefract: <a href="https://catalog.librivox.org/people_public.php?peopleid=2825">Ernst Pattynama</a><br>Lord Alfred Rufford: <a href="https://catalog.librivox.org/people_public.php?peopleid=6446">Bob Gonzalez</a><br>Mr. Kelvil, M.P.: <a href="https://catalog.librivox.org/people_public.php?peopleid=6459">Nigel Boydell</a><br>The Ven. Archdeacon Daubeny, D.D. : <a href="https://catalog.librivox.org/people_public.php?peopleid=3595">Martin Geeson</a><br>Gerald Arbuthnot: <a href="https://catalog.librivox.org/people_public.php?peopleid=1492">mb</a><br>Farquhar, Butler: <a href="https://catalog.librivox.org/people_public.php?peopleid=6482">Delmar H Dolbier</a><br>Francis, Footman: <a href="https://catalog.librivox.org/people_public.php?peopleid=3664">Barry Eads</a><br>Lady Hunstanton: <a href="https://catalog.librivox.org/people_public.php?peopleid=5265">Susanna</a><br>Lady Caroline Pontefract: <a href="https://catalog.librivox.org/people_public.php?peopleid=4964">Rashada</a><br>Lady Stutfield: <a href="https://catalog.librivox.org/people_public.php?peopleid=6281">Beth Thomas</a><br>Mrs. Allonby: <a href="https://catalog.librivox.org/people_public.php?peopleid=1259">Elizabeth Klett</a><br>Miss Hester Worsley: <a href="https://catalog.librivox.org/people_public.php?peopleid=3157">TriciaG</a><br>Alice, Maid: <a href="https://catalog.librivox.org/people_public.php?peopleid=3816">Bev J. Stevens</a><br>Mrs. Arbuthnot: <a href="https://catalog.librivox.org/people_public.php?peopleid=3536">Arielle Lipshaw</a><br>Narrator: <a href="https://catalog.librivox.org/people_public.php?peopleid=2911">David Lawrence</a><br><br><strong>Audio edited by: </strong>Arielle Lipshaw<br>

24Picture Of Dorian Gray (1891 Version)

By Oscar Wilde

The Picture of Dorian Gray is the only published novel by Oscar Wilde, appearing as the lead story in Lippincott's Monthly Magazine on 20 June 1890, printed as the July 1890 issue of this magazine. Wilde later revised this edition, making several alterations, and adding new chapters; the amended version was published by Ward, Lock, and Company in April 1891. The novel tells of a young man named Dorian Gray, the subject of a painting by artist Basil Hallward. Basil is impressed by Dorian's beauty and becomes infatuated with him, believing his beauty is responsible for a new mode in his art. Dorian meets Lord Henry Wotton, a friend of Basil's, and becomes enthralled by Lord Henry's world view. Espousing a new hedonism, Lord Henry suggests the only things worth pursuing in life are beauty and fulfillment of the senses. Realizing that one day his beauty will fade, Dorian (whimsically) expresses a desire to sell his soul to ensure the portrait Basil has painted would age rather than himself. Dorian's wish is fulfilled, plunging him into debauched acts. The portrait serves as a reminder of the effect each act has upon his soul, with each sin displayed as a disfigurement of his form, or through a sign of aging. The Picture of Dorian Gray is considered a work of classic gothic fiction with a strong Faustian theme. (Summary by Wikipedia)