Downloads & Free Reading Options - Results

Hi. Jason Pargin (@JohnDiesattheEn) joins Katy and Cody to discuss Ye's disturbing antisemitic comments, Tulsi Gabbard leaving the Democratic Party (say it ain't so!), and the media's insistence on pretending that drug dealers are sneaking rainbow fentanyl into trick-or-treat candy.Support us on our PATREON: http://patreon.com/somemorenewsCheck out our MERCH STORE: https://www.teepublic.com/stores/somemorenews?ref_id=9949SUBSCRIBE to SOME MORE NEWS: https://tinyurl.com/ybfx89rhPlease fill out our SURVEY: https://kastmedia.com/survey/Subscribe to the Even More News and SMN audio podcasts here -Apple Podcasts: https://podcasts.apple.com/us/podcast/some-more-news/id1364825229Spotify: https://open.spotify.com/show/6ebqegozpFt9hY2WJ7TDiA?si=5keGjCe5SxejFN1XkQlZ3w&dl_branch=1Stitcher: https://www.stitcher.com/show/even-more-news Follow us on social media: Twitter: https://twitter.com/SomeMoreNews Instagram: https://www.instagram.com/SomeMoreNews/ Facebook: https://www.facebook.com/SomeMoreNews/ TikTok: https://www.tiktok.com/@somemorenews Make sure your online activity and data is protected with the best VPN money can buy. Visit https://ExpressVPN.com/morenews right now and get three extra months free. Get a 4-week trial, free postage, and a digital scale at https://www.stamps.com/MORENEWS. Thanks to Stamps.com for sponsoring the show!Athletic Greens is going to give you an immune supporting FREE 1-year supply of Vitamin D AND 5 free travel packs with your first purchase if you visit https://athleticgreens.com/morenews today. Support the show!: http://patreon.com.com/somemorenewsSee omnystudio.com/listener for privacy information.

The talk will consist of two parts. In the first part, I will present prior work aimed at developing new algorithms for materials science applications. I developed a novel numerical method to optimize the shape of mechanical parts and maximize their strength. This optimization problem was extremely challenging as it involved changing the topology and therefore classical optimization methods do not apply. A similar idea can be applied to model shape memory alloys and plasticity in metals. In this project, the optimization led to the construction of a tree of laminated microstructures. Constraints from mechanical equilibrium were used to construct an adaptive tree and to determine the optimal number of levels. The second part involves more recent work. I will present the implementation of a new heat flux algorithm to understand failures in micro-devices in LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator); LAMMPS is a high performance code which was benchmarked on Blue Gene L with 90% parallel efficiency. Also I am currently extending the capabilities of ParaDiS (Parallel Dislocation Simulator), another high performance code which has run on 135k processors of Blue Gene L. ©2008 Microsoft Corporation. All rights reserved.

YouTube's algorithms are abhorrent in a new way, details on the next Playstation console, Google makes nice with developers, and three interesting sides to the TikTok story.Sponsors:EveryKey.com 20% off with promocode RIDE20 Mealime AppLinks: YouTube shows 9/11 link on live videos of unrelated Notre Dame fire (The Verge)EXCLUSIVE: WHAT TO EXPECT FROM SONY'S NEXT-GEN PLAYSTATION (Wired) 15 MONTHS OF FRESH HELL INSIDE FACEBOOK (Wired) Next major macOS version will include standalone Music, Podcasts, and TV apps, Books app gets major redesign (9to5Mac) Improving the update process with your feedback (Android Developer's Blog) AT&T sells Hulu stake, valuing the video streaming company at $15 billion (CNBC) Short video app TikTok extends reach in global markets, sees in-app purchases surge more than 200 per cent (South China Morning Post)TikTok downloads ordered to be blocked on iOS and Android in India over porn and other illegal content (TechCrunch) TikTok sensation Lil Nas X rewrites the rules of country music (Axios)

Do you ever long for the good old days? Nostalgia can be such a wonderful thing. We begin this episode with a look at why the past can seem so idyllic and wonderful and what the positive effects of looking back are. http://science.howstuffworks.com/life/inside-the-mind/human-brain/nostalgia4.htmDo you really understand what algorithms are and how they work? You probably should because algorithms are used so often to influence us in making decisions on what to buy, what movies to watch or who to date. Kartik Hosanagar is a professor of technology, digital business and marketing at the Wharton School at the University of Pennsylvania and author of the book A Human's Guide to Machine Intelligence: How Algorithms Are Shaping Our Lives and How We Can Stay in control (https://amzn.to/2JuStcy). Listen and hear the good and the bad about how algorithms have become part of everyday life."Sit up straight and don't slouch!" How many times have you heard that or said that to someone else? It turns out that posture is really important and I'll explain why. http://www.dailymail.co.uk/health/article-2295420/Stand-straight-stay-fightingfit-From-raised-blood-pressure-bloated-stomach-surprising-effectsbad-posture.html#ixzz2OBJ72FZQSome ideas really stick and others quickly fade away. Why? What makes an idea sticky? Chip Heath, author of the book Made to Stick : Why Some Ideas Thrive and Others Die (https://amzn.to/2Y9YaQu) joins me to talk about his research into what makes a really good idea resonate with people. The advice he gives is very practical and will help you create ideas that people will fall in love with. This Week's Sponsors-Skillshare. For 2 months free access to over 25,000 classes go to www.Skillshare.com/something-ADT. For more information on smart home security go to www.ADT.com -Capital One. Visit www.CapitalOne.com What's in your wallet? Learn more about your ad choices. Visit podcastchoices.com/adchoices

This study aims to assess the impact of skin tone biases in algorithms on technology use. Algorithmic bias, an algorithm that deviates from some standard in a manner that systematically dis/advantages a category of users or a category of data over another, in ways that are not designed as a function of that algorithm, is a frequent problem in society. Specifically, many image recognition algorithms historically and currently advantage users with lighter skin tones over users with darker skin tones, causes functional discrepancies in how the technology functions between use. Therefore, this study seeks to see if this bias impacts how people of varying skin tones perceive and use technology. It is important to note that the relationships under investigation here are not inherently driven by skin tone, group membership, or any other feature of the user (i.e., there is nothing inherent about skin tone that would affect technology use). Instead, the proposed relationships reflect the concern over significant social consequences when algorithms are affected by bias.

Do personalization algorithms serve up ever more extreme videos to users, resulting in affective and issue polarization? Based in part on vivid anecdotes, a media narrative has taken hold ascribing worrisome power to hidden algorithms. Academic researchers have struggled to add evidence to this debate due to both data limitations and serious research design challenges. We propose a novel experimental approach that seeks to resolve the impasse on this critical question by randomizing the recommendations themselves. Using a custom-built interface that serves sequences of videos from YouTube's API, we create our own variations on the existing recommendation algorithm by manipulating the ideological balance of recommended video content to randomly assigned groups of subjects. This approach enables an attractive combination of experimental control and external validity without the need for reverse-engineering YouTube's blackbox recommendations. Subjects in our study will choose to watch a sequence of short videos and follow recommendations while the interface tracks watch time and other measures of engagement. After the viewing session, participants will be surveyed on issue opinions, affective polarization, and perceptions of media hostility. This pre-analysis plan describes a three-wave study focusing on the issue of gun control.

Do personalization algorithms serve up ever more extreme videos to users, resulting in affective and issue polarization? Based in part on vivid anecdotes, a media narrative has taken hold ascribing worrisome power to hidden algorithms. Academic researchers have struggled to add evidence to this debate due to both data limitations and serious research design challenges. We propose a novel experimental approach that seeks to resolve the impasse on this critical question by randomizing the recommendations themselves. Using a custom-built interface that serves sequences of videos from YouTube's API, we create our own variations on the existing recommendation algorithm by manipulating the ideological balance of recommended video content to randomly assigned groups of subjects. This approach enables an attractive combination of experimental control and external validity without the need for reverse-engineering YouTube's blackbox recommendations. Subjects in our study will choose to watch a sequence of short videos and follow recommendations while the interface tracks watch time and other measures of engagement. After the viewing session, participants will be surveyed on issue opinions, affective polarization, and perceptions of media hostility. This pre-analysis plan describes a three-wave study focusing on the issue of gun control.

Do personalization algorithms serve up ever more extreme videos to users, resulting in affective and issue polarization? Based in part on vivid anecdotes, a media narrative has taken hold ascribing worrisome power to hidden algorithms. Academic researchers have struggled to add evidence to this debate due to both data limitations and serious research design challenges. We propose a novel experimental approach that seeks to resolve the impasse on this critical question by randomizing the recommendations themselves. Using a custom-built interface that serves sequences of videos from YouTube's API, we create our own variations on the existing recommendation algorithm by manipulating the ideological balance of recommended video content to randomly assigned groups of subjects. This approach enables an attractive combination of experimental control and external validity without the need for reverse-engineering YouTube's blackbox recommendations. Subjects in our study will choose to watch a sequence of short videos and follow recommendations while the interface tracks watch time and other measures of engagement. After the viewing session, participants will be surveyed on issue opinions, affective polarization, and perceptions of media hostility. This pre-analysis plan describes a three-wave study focusing on the issue of gun control.

In this dissertation, a series of studies are designed to investigate the brain connectivity patterns (both functional connectivity and effective connectivity), comparing subjects with depression history vs. control and subjects with OCD vs. control. Depression and OCD are examples of trans-diagnostic disorders, which may be more likely due to alteration of connectivity patterns (ex., hyper-connectivity in default mode network or other networks) rather than localized brain abnormality.

This study investigates consumers' search and purchase behavior on the Amazon.it marketplace, with a special focus on whether and how they rely on the "Buy-Box". Across treatments, we manipulate the amount of information disclosed to consumers about the functioning (and the hidden risks) of this peculiar platform feature, and observe their behavior on the online marketplace, holding fixed the current design of the platform. We further manipulate the characteristics of the (exogenously determined) products for which we ask participants to perform the e-shopping task.

We suggest that people develop an implicit bias (i.e., prejudice) against artificial intelligence (AI) systems, as a different and threatening “species” the behavior of which is unknown. We expected people to be guided by this implicit bias but try to override it. This leads to some willingness to rely on algorithmic advice (appreciation) which is reduced as a function of people’s implicit prejudice against the machine. Next, we provide an explanation as to why aversion is ephemeral. As people learn about the performance of an algorithm, they depend less on primal implicit biases when deciding whether to rely on the AI’s advice. Two studies (n1=675, n2=317) that use the Implicit Association Test consistently support this view. Two additional studies (n3=255, n4=332) rule out alternative explanations and provide stronger support for our assertions.

Brian interviews two different guests.First is Hillary Boehm-Edwards. Hillary is a first-year commerce student at McMaster University. She talked about an article she wrote and published in the Toronto Star, an opinion editorial regarding the dangers of media algorithms and their ability to polarize opinions. Hillary explained how media algorithms influence the content users see on social media, especially within a political context and elaborates on how these algorithms contribute to misinformation and vaccine hesitancy? The article raises the question of who controls a user's opinions - the viewer or the algorithm. Could you discuss the balance of responsibility between the two?Then, Jenny Kim talks about the Kindfulness Workshop with the Urban Buddhist Monk's 6th Annual Kindful Canada Symposium 2023 event which happens this Saturday, Oct. 14th.

We welcome Kelvin and Vincent to the show to discuss Data Structures and Algorithms in Swift.

In this study we gathered extracted information from different databases and studied the performance of machine learning algorithms in detecting skull base tumors via CT & MRI

Acute coma, a life-threatening condition often encountered in emergency medical settings, represents a major challenge in clinical neurology due to its heterogeneous causes and time-sensitive nature. The condition results from various disruptions in neuronal functions, ranging from primary central nervous system (CNS) diseases to diffuse physiological dysfunctions. Despite the advances in neuroimaging, laboratory diagnostics, and clinical evaluations, early identification and effective management of acute coma remain difficult. Current data indicate that disorders of consciousness account for approximately 10% of neurological illnesses, with acute coma affecting around 5% of patients presenting with consciousness disturbances in the emergency department. The mortality rate in these cases is alarmingly high, reaching up to 30%, further underscoring the urgency of accurate diagnosis and intervention.

The aim of this scoping review is to gather together studies that seek to improve the accuracy of dispatch system to allocate appropriate resources to the injured patients, whether computer-aided dispatch, staff-lead phone call systems, non-computer aided (e.g. flip-card) systems. This includes the identification of patients at high or low risk of adverse outcomes either in the prehospital phase on after arrival at hospital. These studies will then be evaluated with in clinical prediction rule framework.

Unique games are constraint satisfaction problems that can be viewed as a generalization of MAX CUT to a larger domain: We are given a graph G = (V,E) on n vertices and a permutation P uv on the set of labels {1,...,k} for every edge (u, v). Our goal is to assign a label X u in {1,..., k} to each vertex u, so as to maximize the number of satisfied constraints P uv (X u ) = X v . This problem has recently attracted a lot of attention since hardness of approximation for many problems, such as Sparsest Cut and Vertex Cover, was proved assuming the Unique Games Conjecture. Roughly speaking, this conjecture says that even if almost all constraints in a unique game are satisfiable it is NP-hard to satisfy a small constant fraction of constraints.Unique games pose a great challenge for our existing techniques:Typically, semidefinite programming (SDP) relaxations are well suited for optimization problems involving boolean variables (e.g. MAX CUT). But little is known about how to analyze SDP solutions for problems with larger domains. We present three approximation algorithms for Unique Games that satisfy roughly k -epsilon/2 , 1 - O(sqrt{epsilon log k}) and 1 - epsilon * O(sqrt{log k log n}) fraction of all constraints if a (1-epsilon) fraction of all constraints is satisfiable. This talk is based on joint papers with Moses Charikar, Eden Chlamtac, and Konstantin Makarychev. ©2007 Microsoft Corporation. All rights reserved.

Even commercial router vendors have adopted randomized algorithms in a few cases because of their simplicity, speed, and memory-efficiency in a few cases. Further, because of the opportunity to see every member of population (i.e., every arriving packet) and preserve summary information about the entire population, such randomized algorithms can obtain an "edge" over standard algorithms that merely sample the population. I illustrate this thesis by three algorithms. First, I will describe a simple algorithm for finding sources that send a large proportion of traffic, and its application in a worm detection chip. Second, I will describe an algorithm that provides an inexpensive technique for measuring the average and variance of packet latencies and loss on a link. By contrast, the majority of routers have no support for fine-grained latency measurement; managers must instead rely on approximate methods such as sending probe packets or using "tomographic" techniques. If time permits, I will describe a third algorithm which allows scalable logging, say of millions of network addresses infected after an attack, using only a small amount of memory. In all three cases I will quantify the edge obtained over simple sampling. ©2010 Microsoft Corporation. All rights reserved.

The Microsoft Biology Initiative (MBI) is an effort in Microsoft Research to bring new technology and tools to the area of bioinformatics and biology. This initiative is comprised of two primary components, the Microsoft Biology Foundation (MBF) and the Microsoft Biology Tools (MBT). The Microsoft Biology Foundation (MBF) is a language-neutral bioinformatics toolkit built as an extension to the Microsoft .NET Framework, initially aimed at the area of Genomics research. Currently, it implements a range of parsers for common bioinformatics file formats; a range of algorithms for manipulating DNA, RNA, and protein sequences; and a set of connectors to biological web services such as NCBI BLAST. MBF is available under an open source license, and executables, source code, demo applications, and documentation are freely downloadable. The Microsoft Biology Tools (MBT) are a collection of tools targeted at helping the biology and bioinformatics researcher be more productive in making scientific discoveries. The tools provided here take advantage of the capabilities provided in the Microsoft Biology Foundation, and are good examples of how MBF can be used to create other tools. This tutorial will provide an overview of the library, details about how to extend and re-use the library, and demonstrations of the tools released that use the library: The MSR Biology Extension for Excel and the MSR Sequence Assembler. ©2010 Microsoft Corporation. All rights reserved.

In this keynote talk for the Cloud Futures Workshop 2011, Michael Franklin describes the work in the Algorithms, Machines, and People Lab (AMPLab)—an industry-supported collaboration of students, postdocs, and faculty who specialize in data management, cloud computing, statistical machine learning, and other important topics that are necessary for making sense of vast amounts of heterogeneous and unruly data. ©2011 Microsoft Corporation. All rights reserved.

The allocation of processes to processors has long been of interest to engineers. The processor allocation problem considered here assigns multiple applications onto a computing system. With this algorithm researchers could more efficiently examine real-time sensor data like that used by United States Air Force digital signal processing efforts or real-time aerosol hazard detection as examined by the Department of Homeland Security. Different choices for the design of a load balancing algorithm are examined in both the problem and algorithm domains. Evolutionary algorithms are used to find near-optimal solutions. These algorithms incorporate multiobjective coevolutionary and parallel principles to create an effective and efficient algorithm for real-world allocation problems. Three evolutionary algorithms (EA) are developed. The primary algorithm generates a solution to the processor allocation problem. This allocation EA is capable of evaluating objectives in both an aggregate single objective and a Pareto multiobjective manner. The other two EAs are designed for fine turning returned allocation EA solutions. One coevolutionary algorithm is used to optimize the parameters of the allocation algorithm. This meta-EA is parallelized using a coarse-grain approach to improve performance. Experiments are conducted that validate the improved effectiveness of the parallelized algorithm. Pareto multiobjective approach is used to optimize both effectiveness and efficiency objectives. The other coevolutionary algorithm generates difficult allocation problems for testing the capabilities of the allocation EA. The effectiveness of both coevolutionary algorithms for optimizing the allocation EA is examined quantitatively using standard statistical methods. Also the allocation EAs objective tradeoffs are analyzed and compared.

This report presents the user menu for a prototype of the Monitor, Emplacement, and Control in a Halo (MECH) model for modeling and simulation of behaviors associated with insurgent attacks, and their relationship with geographic locations and temporal windows. The software system is consisted of an Android device to make query to a server, which can make tactical analysis using the MECH model based statistical pattern analysis and simulation tools. The statistical pattern analysis method is cross checked with tactical simulations based on the MECH behavior model. The prototype demonstrates the effectiveness of fusion of statistical pattern analysis, simulation, and human interpretation of military doctrines within the context of the two modeling approaches. It shows the feasibility of self-guided situational analysis informed by MECH-based situational awareness analytics.

This study investigated the minimum requirements to establish a satellite tracking system architecture for a microsatellite to rendezvous with a non-cooperative target satellite. A prototype optical tracking system was reviewed with emphasis on a proposed tactical employment that could be used by technologically unsophisticated state or non-state adversaries. With the tracking system architecture selected, simulated tracking data was processed with a Non-Linear Least Squares batch orbit estimation algorithm and a Bayes sequential orbit determination filter to update the target satellite's state vector.

This project was concerned primarily with one central theme which is the attempt to use quantum mechanics to design algorithms that perform better than conventional (non-quantum) algorithms for solving certain problems. We looked at a variety of approaches. The first is quantum adiabatic evolution which was invented by the authors along with M. Sipser and S. Gutmann. During the period of this report we studied the robustness of this algorithm, generalized the algorithm beyond its original specification and also showed that it can outperform classical algorithms in certain settings. We also investigated continuous time quantum walks which were introduced as a quantum algorithmic tool by Farhi and Gutmann. Here the high point was the discovery of a quantum walk algorithm that gives provable exponential speedup over the best possible classical algorithm for a certain oracle problem. Goldstone and Childs (a graduate student at MIT at the time) explored the use of quantum walk algorithms for searching a spatial grid. Also Farhi, Goldstone and coworkers showed how to use repeated measurements as an algorithmic tool. In particular we showed how to achieve the Grover square root speedup using measurement algorithms. In the remainder of the report we will elaborate on these findings and make reference to the associated papers.

In this research program we developed novel analytical and computational methods to estimate the support of radiating sources and scatterers from knowledge of the corresponding far field radiation pattern or scattering amplitude. The focus was the development of methods to estimate the so-called minimum source region or scattering support of a far field. The support of any source that generates the given far field must contain this minimum source region. The results were derived in the framework of the Helmholtz equation. The derived methodology consisted of two components, one for the estimation of convex support information, and a subsequent procedure to estimate the nonconvex support. This project also led to new theoretical and computational developments in estimation-theoretic assessment of radar target separation estimability from far field data as well as to a new change detection method called the optical theorem detector, which is relevant to detect unknown targets and changes in random and complex media.

Talk by Ron Eglash for "The Limits of Language in Computational Cultures" workshop.  Organized by code-anth and unixjazz. April 2022. 

An exercise in lunacy when I got tired of rapping as a frog. In the long hours of work nights, my brain becomes a playground for nonsensical demon thoughts, so naturally, I turn on a microphone. When I wake up the next evening and observe what horrific sounds uttered from my gullet the night before, I feel it absolutely necessary to share them with as many people as possible. Only a sleep-deprived person could produce something like this, and for the sake of your own insanity, everyone needs to soak such windows into dementia in.

This study compares the performance of two estimation algorithms of new usage, the Metropolis-Hastings Robins-Monro (MHRM) and the Hamiltonian MCMC (HMC), with two consolidated algorithms in the psychometric literature, the marginal likelihood via EM algorithm (MML-EM) and the Markov chain Monte Carlo (MCMC), in the estimation of multidimensional item response models of various levels of complexity. This paper evaluates the performance of parameter recovery via three simulation studies from a Bayesian approach. The first simulation uses a very simple unidimensional model to evaluate the effect of diffuse and concentrated prior distributions on recovery. The second study compares the MHRM algorithm with MML-EM and MCMC in the estimation of an item-response model with a moderate number of correlated dimensions. The third simulation evaluates the performance of the MHRM, HMC, MML-EM and MCMC algorithms in the estimation of an item response model in a high-dimensional latent space. The results showed that MML-EM loses precision with high-dimensional models whereas the other three algorithms recover the true parameters with similar precision. Apart from this, the main differences between algorithms are: (1) estimation time is much shorter for MHRM than for the other algorithms; (2) MHRM achieves the best precision in all conditions and is less affected by prior distributions; and (3) prior distributions for the slopes in the MCMC and HMC algorithms should be carefully defined in order to avoid problems of factor orientation. In summary, the new algorithms seem to overcome the difficulties of the traditional ones by converging faster and producing accurate results.

The topic of sorting algorithms is a pleasant subject of informatics education. Not only is it so because the notion of sorting is well known from our everyday life, but also because as an algorithm task, whether we expect naive or practical solutions, it is easy to define and demonstrate. In my paper I will present some of the possible methods and goals of teaching sorting algorithms in the primary and the secondary school context. Some of the applicable demonstration and collaborative methods will be listed and illustrated with examples available on the internet. Next to defining the teaching goals offered by the specific methods, I will also make suggestions which methods are optimal for a given teaching phase.

We study the phenomenon of lack of reversibility in molecular dynamics algorithms for the case of Wilson's lattice QCD. We demonstrate that the classical equations of motion that are employed in these algorithms are chaotic in nature. The leading Liapunov exponent is determined in a range of coupling parameters. We give a quantitative estimate of the consequences of the breakdown of reversibility due to round-off errors.

We discuss two ways in which parton shower algorithms can be supplemented by matrix-element corrections to ensure the correct hard limit: by using complementary phase-space regions, or by modifying the shower itself. In the former case, existing algorithms are self-consistent only if the total correction is small. In the latter case, existing algorithms are never self-consistent, a problem that is particularly severe for angular-ordered parton shower algorithms. We show how to construct self-consistent algorithms in both cases. The postscript file for this paper can also be obtained by anonymous ftp from thep.lu.se in the file pub/Preprints/lu_tp_94_17.ps

We discuss two ways in which parton shower algorithms can be supplemented by matrix-element corrections to ensure the correct hard limit: by using complementary phase-space regions, or by modifying the shower itself. In the former case, existing algorithms are self-consistent only if the total correction is small. In the latter case, existing algorithms are never self-consistent, a problem that is particularly severe for angular-ordered parton shower algorithms. We show how to construct self-consistent algorithms in both cases. The postscript file for this paper can also be obtained by anonymous ftp from thep.lu.se in the file pub/Preprints/lu_tp_94_17.ps

A complete description is given for the algorithms in S+WAVELETS software toolkit for wavelet and cosine packet analysis. These algorithms include wavelet transforms, wavelet packet transforms, cosine packet transforms, and non-decimated wavelet transforms. Implementations for the transforms and their inverses are given for a variety of boundary treatment rules, including periodic, reflection, interval wavelets, and zero/polynomial extension. In addition, modifications to the standard algorithms are given to handle signals or images with dimensions not divisible by a power of two.

In this thesis we propose several algorithms for rank-one updates and downdates to these decompositions with strong stability properties and efficient implementations on high-performance computers. We seek algorithms which only require O(n2) operations per update or downdate unlike recomputing the two-sided orthogonal decomposition (TSOD) in O(n3). We also desire highly regular data movement inherited in these algorithms in order to implement these efficiently on the distributed memory MIMD multiprocessors. The algorithms are based upon 'chasing' strategies for updating and downdating procedures for orthogonal decompositions. (AN)

The aim of this project was to demonstrate that a systems based methodology, utilizing the tools of fast algorithms, signal processing, and multivariable control, can make a significant difference to the efficient solution of various critical problems in the fields of semiconductor manufacturing and materials processing. Specifically, the following developments were achieved: (1) A robust multivariable control algorithm for use in several types of rapid thermal processes including oxidation, annealing, and silicidation as well as single-wafer chemical vapor deposition processes: (2) Introduction of novel signal processing techniques into a recently proposed acoustics based technique for noninvasive temperature measurement of RTP wafers; (3) Subspace based image processing strategies for defect inspection of periodic patterns in patterned wafers, distortion compensation for accurate overlay in lithography of quasiperiodic patterns, circle and ellipse fitting, and critical dimension measurement; and (4) Fast algorithms for the systematic design of phase-shifting masks.

Advances in speech technology and computing power have created a surge of interest in the practical application of speech recognition. However, the most accurate speech recognition systems in the research world are still far too slow and expensive to be used in practical, large vocabulary continuous speech applications. Their main goal has been recognition accuracy, with emphasis on acoustic and language modelling. But practical speech recognition also requires the computation to be carried out in real time within the limited resources CPU power and memory size of commonly available computers. There has been relatively little work in this direction while preserving the accuracy of research systems. In this thesis, we focus on efficient and accurate speech recognition. It is easy to improve recognition speed and reduce memory requirements by trading away accuracy, for example by greater pruning, and using simpler acoustic and language models. It is much harder to improve both the recognition speed and reduce main memory size while preserving the accuracy. This thesis presents several techniques for improving the overall performance of the CMU Sphinx-II system. Sphinx-II employs semi-continuous hidden Markov models for acoustics and trigram language models, and is one of the premier research systems of its kind. The techniques in this thesis are validated on several widely used benchmark test sets using two vocabulary sizes of about 20K and 58K words. The main contributions of this thesis are an 8-fold speedup and 4-fold memory size reduction over the baseline Sphinx-II system. The improvement in speed is obtained from the following techniques: lexical tree search, phonetic fast match heuristic, and global best path search of the word lattice.

Access to communication networks is increasing rapidly. The increase access to these networks results in delays and at times loss of data. At times of peak traffic or when trunks or nodes are down, very important customers' communications requirements are not met. One way to combat this problem is to prioritize the network and provide different levels of grade of service (GOS) for each priority. Call preemption provides an effective method of obtaining different levels of GOS. This research seeks to design the best circuit switch communications network preemption model for the DoD by analyzing previously developed preemption algorithms. Four simulation network models are developed. The grades of service per priority are obtained as the network capacity decreases and as the calls generated in node 0 increases. The analysis of preemption network models is based on the grade of service, average number of preemptions, and average network bandwidth. The networks are simulated under the same input parameters. The analysis showed that preemption can significantly lower the grade of service for high priority customers in a congested network. The best configuration preemption model depends on the bandwidth flexibility of the network and the goals of the communications network organization.

11 p. ; 28 cm

The Computer Aided Prototyping System (CAPS) and the Prototype system Description Language (PSDL) represent a pioneering effort in the field of software development. Execution Support System (ESS) within the framework of CAPS. The Static Scheduler is one of the critical elements of the ESS which extracts critical timing constraints and precedence constraints for operators and schedules the time-critical operators to guarantee that their timing constraints will be met. The Static Scheduler uses the information of timing constraints and precedence constraints to determine whether a feasible schedule can be built. This construction provides the foundation for handling the execution for Real-Time systems. The goal of this thesis is to provide improved versions of the Static Scheduler.

In this work we propose and numerically analyze an algorithm for detection of a contaminant source using a dynamic sensor network. The algorithm is motivated using a global probabilistic optimization problem and is based on the analysis of the hydrodynamic limit of a discrete time evolution equation on the lattice under a suitable scaling of time and space. Numerical results illustrating the effectiveness of the algorithm are presented.

Sequential Monte Carlo algorithms, or Particle Filters, are Bayesian filtering algorithms which propagate in time a discrete and random approximation of the a posteriori distribution of interest. Such algorithms are based on Importance Sampling with a bootstrap resampling step which aims at struggling against weights degeneracy. However, in some situations (informative measurements, high dimensional model), the resampling step can prove inefficient. In this paper, we revisit the fundamental resampling mechanism which leads us back to Rubin's static resampling mechanism. We propose an alternative rejuvenation scheme in which the resampled particles share the same marginal distribution as in the classical setup, but are now independent. This set of independent particles provides a new alternative to compute a moment of the target distribution and the resulting estimate is analyzed through a CLT. We next adapt our results to the dynamic case and propose a particle filtering algorithm based on independent resampling. This algorithm can be seen as a particular auxiliary particle filter algorithm with a relevant choice of the first-stage weights and instrumental distributions. Finally we validate our results via simulations which carefully take into account the computational budget.

Consider a transmission scheme with a single transmitter and multiple receivers over a faulty broadcast channel. For each receiver, the transmitter has a unique infinite stream of packets, and its goal is to deliver them at the highest throughput possible. While such multiple-unicast models are unsolved in general, several network coding based schemes were suggested. In such schemes, the transmitter can either send an uncoded packet, or a coded packet which is a function of a few packets. The packets sent can be received by the designated receiver (with some probability) or heard and stored by other receivers. Two functional modes are considered; the first presumes that the storage time is unlimited, while in the second it is limited by a given Time to Expire (TTE) parameter. We model the transmission process as an infinite-horizon Markov Decision Process (MDP). Since the large state space renders exact solutions computationally impractical, we introduce policy restricted and induced MDPs with significantly reduced state space, and prove that with proper reward function they have equal optimal value function (hence equal optimal throughput). We then derive a reinforcement learning algorithm, which learns the optimal policy for the induced MDP. This optimal strategy of the induced MDP, once applied to the policy restricted one, significantly improves over uncoded schemes. Next, we enhance the algorithm by means of analysis of the structural properties of the resulting reward functional. We demonstrate that our method scales well in the number of users, and automatically adapts to the packet loss rates, unknown in advance. In addition, the performance is compared to the recent bound by Wang, which assumes much stronger coding (e.g., intra-session and buffering of coded packets), yet is shown to be comparable.

Presentation Overview: Project Overview; ERDC SUGV testing data analysis; Algorithm Analysis; Terrain EDM Analysis; Performance Comparisons; Future Work; Summary.

Consider the Maximum Weight Independent Set problem for rectangles: given a family of weighted axis-parallel rectangles in the plane, find a maximum-weight subset of non-overlapping rectangles. The problem is notoriously hard both in the approximation and in the parameterized setting. The best known polynomial-time approximation algorithms achieve super-constant approximation ratios [Chalermsook and Chuzhoy, SODA 2009; Chan and Har-Peled, Discrete & Comp. Geometry 2012], even though there is a $(1+\epsilon)$-approximation running in quasi-polynomial time [Adamaszek and Wiese, FOCS 2013; Chuzhoy and Ene, FOCS 2016]. When parameterized by the target size of the solution, the problem is $\mathsf{W}[1]$-hard even in the unweighted setting [Marx, FOCS 2007]. To achieve tractability, we study the following shrinking model: one is allowed to shrink each input rectangle by a multiplicative factor $1-\delta$ for some fixed $\delta>0$, but the performance is still compared against the optimal solution for the original, non-shrunk instance. We prove that in this regime, the problem admits an EPTAS with running time $f(\epsilon,\delta)\cdot n^{\mathcal{O}(1)}$, and an FPT algorithm with running time $f(k,\delta)\cdot n^{\mathcal{O}(1)}$, in the setting where a maximum-weight solution of size at most $k$ is to be computed. This improves and significantly simplifies a PTAS given earlier for this problem [Adamaszek et al., APPROX 2015], and provides the first parameterized results for the shrinking model. Furthermore, we explore kernelization in the shrinking model, by giving efficient kernelization procedures for several variants of the problem when the input rectangles are squares.

Generating a defined set of genetic constructs within a large combinatorial space provides a powerful method for engineering novel biological functions. However, the process of assembling more than a few specific DNA sequences can be costly, time consuming and error prone. Even if a correct theoretical construction scheme is developed manually, it is likely to be suboptimal by any number of cost metrics. Modular, robust and formal approaches are needed for exploring these vast design spaces. By automating the design of DNA fabrication schemes using computational algorithms we can eliminate human error while reducing redundant operations, thus minimizing the time and cost required for conducting biological engineering experiments. Here, we provide algorithms that optimize the simultaneous assembly of a collection of related DNA sequences. We compare our algorithms to an exhaustive search on a small synthetic dataset and our results show that our algorithms can quickly find an optimal solution. Comparison with random search approaches on two real-world datasets show that our algorithms can also quickly find lower-cost solutions for large datasets.

For military systems, environmental stressors (e.g. motion, temperature, noise) must be considered during decision making related to manpower requirements, workload determination, design tradeoffs, and mission effectiveness/sustainability early into and throughout the system acquisition process. Current human performance modeling techniques may have limited predictive utility and have not been fully validated against operational human in the loop (HIL) data. As a result, they may lack sufficient fidelity to support systems engineering needs to predict the individual and interactive effects that environmental stressors may have on human performance. The purpose of this paper is to describe an approach for developing performance shaping function (PSF) algorithms for environmental stressors that can be integrated into human performance modeling tools. These high fidelity plug-in algorithms are anticipated to provide an enhanced level of predictive validity when compared to current discrete event modeling tools. The algorithms will address environmentally induced limitations that are levied on human performance and enhance decision making in defense acquisition system design and cost versus performance tradeoffs.

Applied novel mathematical techniques to a published data-set (Roitman and Shadlen, 2002) on LIP neuronal activities during a random-dot motion discrimination paradigm to identify/classify individual neurons in terms of their sensori-motor locus and the putative decisional process that translates a perceptual to a motor representation. Developed a Signal-Detection-Theory based analysis providing a quantitative measure of sensorimotor locus of the neuron at each time point, and a Poisson regression model incorporating orthogonal decomposition of neuronal activity in term of how the stimulus, response, and stimulus-response mapping components contribute to the spike activity. Developed techniques to extract S-, R-, or SR mapping components in neural recordings of time series, e.g. event-related potentials (ERPs) where trial-by-trial variations in response-time have contaminated their contributions in averaged waveforms. Modeled motivational impact ( incentive salience ) in reinforcement learning, a standard paradigm for sequential decision making.

This paper addresses the problem of target detection and classification, where the performance is often limited due to high rates of false alarm and classification error, possibly because of inadequacies in the underlying algorithms of feature extraction from sensory data and subsequent pattern classification. In this paper, a recently reported feature extraction algorithm, symbolic dynamic filtering (SDF), is investigated for target detection and classification by using unmanned ground sensors (UGS). In SDF, sensor time series data are first symbolized to construct probabilistic finite state automata (PFSA) that, in turn, generate low-dimensional feature vectors. In this paper, the performance of SDF is compared with that of two commonly used feature extractors, namely Cepstrum and principal component analysis (PCA), for target detection and classification. Three different pattern classifiers have been employed to compare the performance of the three feature extractors for target detection and human/animal classification by UGS systems based on two sets of field data that consist of passive infrared (PIR) and seismic sensors. The results show consistently superior performance of SDF-based feature extraction over Cepstrum-based and PCA-based feature extraction in terms of successful detection, false alarm, and misclassification rates.

The project has produced new computational models and algorithms for coordination, prediction and planning in situations involving multiple decision makers that operate over an extended period of time in either collaborative or adversarial domains. This includes the development of the decentralized partially-observable Markov decision process (DEC-POMDP) model, memory-bounded algorithm for solving finite-horizon DEC-POMDPs, sparse representations of agent strategies using finite-state controllers, bounded policy iteration algorithms for infinite-horizon DEC-POMDPs, and algorithms for solving DEC-POMDPs using non-linear optimization methods. The project produced the best existing exact algorithms for these problems as well as scalable approximation techniques and benchmark problems that are now widely used within the multi-agent systems community. The report describes these research accomplishments and provides references to published papers and PhD dissertations that include detailed descriptions of the results.

In this paper, we present a methodology to assess the results of image processing algorithms for unstructured road edges detection and tracking. We aim at performing a quantitative, comparative and repetitive evaluation of numerous algorithms in order to direct our future developments in navigation algorithms for military unmanned vehicles. The main scope of this paper is the constitution of this database and the definition of the assessment metrics.