Program Development Study Phase - Gsa: Fill & Download for Free

Q: How well do the US Army Special Forces learn to speak their assigned foreign language (assuming no previous knowledge of the language)?Overall, U.S. Army Special Forces (SF) learn to speak their assigned foreign language by attending the various military & DOD language schools/training offered. High-level language skills for SF personnel are still a requirement for current & future operations, but they are very difficult and costly to achieve & maintain. As they said to us before any class attended, “foreign language proficiency is a perishable skill” which means use it or lose it.In response to the needs for high-level foreign language in SF units, the United States Army Special Forces Command (USASFC) set the goal of having at least one SF team member qualified at the 2/2 & 3/3 levels in each deploying twelve-man Operational Detachment Alpha (ODA). The Defense Language Institute (DLI) considers a 2/2 to be “Limited Working Proficiency” & is the minimum score for certification as a Defense Department Basic Linguist while a 3/3 is considered “General Professional Proficiency” & is the score needed for most positions in the Defense Attaché system.It has been estimated the hours & what is achieved between the initial foreign language training at the Special Forces Qualification Course (SFQC) & the language courses at the Defense Language Institute (DLI). Language training at the SFQC is only designed to train students to the 1/1 level (elementary proficiency), while the DLI advanced courses produce 3/3 speakers & the DLI basic courses produce 2/2 speakers. There is a gap between the SFQC & DLI as an approximation of the additional training Special Forces Soldiers require to reach advanced proficiencies. This gap was multiplied by the hourly rate for foreign language instruction at a major General Services Administration (GSA) certified foreign language contractor to determine the cost required to achieve proficiency at the 2/2 and 3/3 level in all SOF priority languages. A total cost for the program was then determined under four language distribution scenarios: a base scenario that evenly distributed languages in each SF group, a current needs scenario that placed priority on operationally critical languages such as Arabic & Chinese, a low-cost scenario that assigned all the easiest to learn language & least expensive, relevant to their assigned SF group, and a high-cost scenario that assigned all the hardest to learn language and therefore most expensive relevant to their assigned SF group.Foreign Language skills are key for current and future SF missions for Army SF. Army SF is often confused with the broader term Special Operations Forces (SOF). SOF refers to all forces assigned to US Special Operations Command & includes not only SF units but also Navy SEALS, Air Force Commandos, Marine Corps Special Operations personnel as well as other Army Special Operations units such as the 75th Ranger Regiment & the 160th Special Operations Aviation Regiment. Army SF is the Defense Department’s DOD) primary force for conducting unconventional warfare (UW) & foreign internal defense (FID) along with the 95th Civil Affairs Brigade & the 4th Military Information Support Group, are distinct from other SOF units due in large part to their language training & cultural and regional expertise. Active duty SF qualified Soldiers are assigned to one of five regionally oriented SF groups where the individual language & cultural skills of these SF are leveraged to execute the regional specific missions of each SF group. As an example, an SF qualified Soldier initially assigned to the PACOM oriented 1st SF Group would receive language training in an Asian language such as Chinese, Korean, Thai, Tagalog or Indonesian with some being harder or easier to learn than others.Many years of post-9/11 operational experience have reinforced what SOF leaders have long understood which is foreign language skills are essential to conducting the unique missions of Army SF. Going back to their World War II roots in the OSS, SF has long been distinguished by their foreign language skills. In recent years, these skills have only become more important. In the foreword to the Army’s doctrinal publication involving SF, ADP 3-05, Special Operations, it was written that success for SOF in future conflicts requires a profound understanding of foreign culture & fluency in local languages. It has been stressed the operational need for SOF personnel with languages, more cultural attunements, & regional expertise.USASFC set ambitious goals for its organic foreign language capabilities in response to the need for high-level foreign language skills for SF personnel. Beyond the basic requirement for all SF to maintain a 1/1 in their assigned languages, USASFC set as the unit objective for each deploying ODA as having one SF proficient at the 2/2 level & one SF proficient at the 3/3 level. These goals continue to be ambitious given that the required minimum for SF is 1/1.Reviewing these requirements to every active duty ODA will require at most 360 2/2 qualified SF & 360 3/3 qualified SF. This number is based on a fully manned SF organization of 4 battalions per SF group, 3 companies per battalion & 6 ODAs per company.SF are provided with 2 types of language training: acquisition and sustainment. Currently, acquisition training comes in the form of Basic Special Operations Language Training (BSOLT). All students in the SFQC attend BSOLT as the last phase of the qualification course before graduation. BSOLT is 26 weeks for all students regardless of the difficulty of the language studied & students are required to achieve a 1/1 before they can earn their green beret. Foreign languages are grouped into 4 categories based on their difficulty. Category I languages, such as French or Spanish, & Category II languages, such as Indonesian, are the easiest to learn, while Category III languages, such as Russian, & Category IV languages, such as Chinese, are the most difficult. SF students are assigned their target language based on their aptitude for learning foreign languages as demonstrated through scores on the Defense Language Aptitude Battery (DLAB).After graduation, SF personnel continue to receive sustainment ( use it or lose it) training & are expected to maintain a 1/1 in their designated target language. Soldiers who fail to maintain a 1/1 are not eligible for advanced skills training & are less competitive for promotion. SF personnel who are assigned a Category I or II language must have, at a minimum, 80 hours of annual foreign language sustainment training while others who have a Category III or IV language must have 120 hours. In all cases only 40 hours of this annual training can be independent study, the remainder must be instructor-led in some form. Based on operational needs & availability, SF personnel also have access to alternative training venues to include distance education through video conferencing, resident courses at DLI & participating universities, intermediate level courses specifically designed for SOF & short-term overseas immersion programs.An analysis of the current SF foreign language program indicates that it will be almost impossible to reach the goal of a 2/2 and 3/3 qualified Solider per ODA without more additional changes to the training language program. Using training at DLI as the standard for the length of time required to achieve high-level proficiency, it is clear that there is an enormous gap between a typical SF Soldier’s initial language training & what is required to achieve a 3/3 or even only a 2/2 in all but the easiest to learn languages.At DLI, the basic course in each language is designed to reach a 2/2 level, the intermediate course a 2+/2+ & the advanced course a 3/3. Graduating the DLI basic course & earning a 2/2 takes between 26 weeks for Category I languages & 64 weeks for Category IV languages. If the DLI basic, intermediate & advanced courses were placed back to back to back, reaching a 3/3 would require 62 weeks for Category I languages & 158 weeks for Category IV languages. As BSOLT is only 26 weeks in length, there is a huge gap between initial acquisition & the DLI requirements for advanced proficiency. For Category, I languages BSOLT is the same length as the DLI basic course, as a result, it is likely that enough learners can earn a 2/2 out of BSOLT to meet USASFC’s goal of a 2/2 speaker per ODA. For the hardest to learn languages there is a gap of up to 38 additional weeks between the length of BSOLT & the DLI Basic Course. For achieving a 3/3, Category I languages have a gap of 36 weeks & Category IV languages have a gap of 132 weeks.Scholarship on foreign language maintenance indicates that it is very difficult for learners with low initial acquisition levels to significantly improve their skills without another iteration of prolonged intensive study. In fact, according to a study by instructors at the State Department’s Foreign Service Institute (FSI), learners who do not initially acquire at least a 3/3 level of proficiency are unlikely to improve & are likely to see their skills degrade outside of intensive formal training. This is significant for SF Soldiers because it will be very difficult for them to improve their skills solely through independent study. Critically, however, if SF personnel achieve very high levels of proficiency initially they are much more likely to retain those skills throughout their careers.Assuming that annual sustainment training will improve an SF’s language score over time, the gap between initial acquisition & a 2/2 or 3/3 level of proficiency is so great that it would take decades to make up the difference. Assuming an SF only received the minimum annual language training & his skills increased at a constant rate, it would still take 13.5 years to reach a 3/3 in a language such as French & it would take over 30 years to do so in a Category IV language like Arabic!!Unfortunately, DOD studies indicate that additional sustainment training must be intensive & lengthy to successfully raise an individual’s proficiency level. Short or part-time command programs will help sustain a skill but are not effective enough to raise a proficiency score. Only full-time training seems to be truly sufficient to raise an SF’s score. Most of the additional training venues available to SF, such as online training or study at participating universities are not long or intensive enough to raise a Soldier’s proficiency level. Even overseas language immersion is not as effective or cost-efficient as intensive daily training for acquiring higher-level langue skills. The learning center for SF personnel, the United States Army John F. Kennedy Special Warfare Center & School (USAJFKSWCS), does offer intensive 32-week intermediate level SOLT where personnel are expected to graduate with a 2/2, but not all languages are currently offered & class sizes & numbers are not sufficient to meet USASFC’s goalIf regular annual training and the additional training venues currently available to Special Forces Soldiers are not sufficient to meet USASFC’s unit goal of a 2/2 and 3/3 qualified Soldier per ODA, then what methods would be sufficient? As described earlier, only long-term, intensive training can reliably improve foreign language proficiency. Though attendance at DLI or additional advanced and intermediate level SOLT classes would be optimal, this strategy would require support from outside agencies or lengthy implementation timelines. The simplest and most direct solution is for USASFC to take recently qualified Special Forces Soldiers and contract a private company to provide enough additional hours of instruction to make up the gap between SOLT and the DLI basic or advanced course. USASFC could select from many GSA approved contractors that already have the necessary capacity to train 720 Soldiers in any of the Special Operations priority languages. This solution would provide the intensive, long-term training that Defense Department studies and outside scholarship indicate are necessary to achieve high-level proficiency.Each SF group has associated languages based on each group’s regional orientation. Because teaching cost increases as languages become more difficult to learn, the number of SF Soldiers studying each language has a significant impact. Here are some things to consider.1) A baseline scenario that evenly distributed languages within each SF Group. For instance, in the CENTCOM oriented 5th SF group, half of the Soldiers were assumed to have Arabic as a target language & half were assumed to have Persian-Farsi.2) A current needs scenario that distributed languages based on current operational priorities. For example, in the PACOM oriented 1st SF Group, half of the soldiers were assumed to have Chinese as their target language.3) A low-cost scenario that analyzed each group with all soldiers assigned the easiest (and therefore least expensive to teach) language associated with their SF group. In this scenario, for example, the EUCOM and AFRICOM oriented 10th SF group as assumed to have all 144 Soldiers learn French4) A high-cost scenario that assumed that all Soldiers would have the most difficult (and therefore most expensive to teach) language in their Special Forces group. For instance, in Central Asia oriented 3rd SF Group, all 144 Soldiers were analyzed with Pasto-Afghan as their target language.In the current fiscal environment, millions of dollars may be too great an expense. However, USASFC could pursue several alternative courses of action that could come close to achieving their language proficiency goal at a significantly lowered cost. In all scenarios, achieving a 3/3 level of proficiency in each ODA constitutes more than 79% of the program’s total cost (that is achieving 2/2 level for 720 Soldiers costs approximately 20% of the program’s total cost, while achieving 3/3 level for the other 720 Soldiers costs approximately 80% of the program’s total cost). As a result, lowering the goal for upper-level proficiency has significant cost-saving opportunities.For example, a less ambitious goal of having a 2+/2+ qualified Soldier (the equivalent of the DLI intermediate course) instead of a 3/3 would reduce the cost of the “current needs” scenario from millions of dollars spent. More dramatic courses of action may be able to achieve the spirit of the USASFC goal of increasing high-level proficiency within deploying ODAs. For instance, because of their relatively greater focus on Special Warfare activities versus Surgical Strike missions, SOF Civil Affairs, & Military Information Support Operations ( MISO) forces may be a more appropriate place for USASOC to develop high-level language skills. Civil Affairs & Miso units regularly deploy with Sf units & a 3/3 speaker in each Civil Affairs Team or Military Information Support Team could provide a similar level of mission effectiveness as having the same level of proficiency organic to the deploying ODA.In conclusion, maybe contracting a private company to provide intensive long-term language training is certainly the most direct & effective way of reaching USASFC’s language goals. This program can be thought of as a major, long-term investment that would meet USASFC’s language needs for possibly as long as ten to fifteen years. Once an initial cohort of 720 Soldiers was qualified at the necessary proficiency level, only a few new Soldiers every year would have to receive high-level language training. Because Soldiers at the 3/3 and 2/2 levels are less likely to lose language proficiency over time, less spending on language sustainment training within the SF would be necessary. As a result, this program may be able to save USASFC money in the long term.

Hardware:Controlling nuclear spin by magnetic fields is a recent innovation which leads to smaller/faster/lower-powered computation.[1]Photon-based quantum computing innovations[2] being developed to allow better upscaling of computations.Software:Equating diophantine equations with automorphic forms[3] makes some problems more tractable to computer solution.There are lots of computer innovations published every day. The following list is from arxiv, searching “computing” showing what was submitted 10 April 2020.arXiv:2004.05151 [pdf] cs.CVModel Uncertainty Quantification for Reliable Deep Vision Structural Health MonitoringAuthors: Seyed Omid Sajedi, Xiao LiangAbstract: Computer vision leveraging deep learning has achieved significant success in the last decade. Despite the promising performance of the existing deep models in the recent literature, the extent of models' reliability remains unknown. Structural health monitoring (SHM) is a crucial task for the safety and sustainability of structures, and thus prediction m… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05148 [pdf, ps, other] hep-thHolographic spin liquids and Lovelock Chern-Simons gravityAuthors: A. D. Gallegos, U. GürsoyAbstract: …Lovelock-Chern-Simons gravity with independent vielbein and spin connection. We construct a hydrodynamic expansion that involves the stress tensor and the spin current and compute the corresponding one-point functions holographically. As a byproduct we find a class of interesting analytic solutions to the Lovelock-Chern-Simons gravity, including blackholes,… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 52 pagesarXiv:2004.05143 [pdf, other] eess.SYResilient State Estimation in Presence of Severe Coordinated Cyber-Attacks on Large-Scale Power SystemsAuthors: Ana Jevtić, Marija IlićAbstract: …based on output clustering. By augmenting the measurement set by respective cluster variables, the system observability is regained, and a reliable state estimate can be computed. We show the numerical performance of our proposed algorithm and its ability to successfully replace corrupted measurements using cluster variables through an example on the IEEE 2… ▽ MoreSubmitted 8 April, 2020; originally announced April 2020.Comments: arXiv admin note: substantial text overlap with arXiv:2004.03838arXiv:2004.05119 [pdf, other] cs.CLSimpleTran: Transferring Pre-Trained Sentence Embeddings for Low Resource Text ClassificationAuthors: Siddhant Garg, Rohit Kumar Sharma, Yingyu LiangAbstract: …or training the classifier and the embedding models end-to-end. Keeping embeddings fixed, SimpleTran significantly improves over fine-tuning on small datasets, with better computational efficiency. With end-to-end training, SimpleTran outperforms fine-tuning on small and medium sized datasets with negligible… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05117 [pdf, other] quant-phA Jordan-Wigner gadget that reduces T count by more than 6x for quantum chemistry applicationsAuthors: Sam PallisterAbstract: Quantum computers have the potential to be a profoundly transformative technology, particularly in the context of quantum chemistry. However, running a chemistry application that is demonstrably useful currently requires a prohibitive number of logical operations. For example, the canonical estimate of the number of operations required to simulate the molecu… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 4 pagesarXiv:2004.05116 [pdf, other] cs.CRA note on blind contact tracing at scale with applications to the COVID-19 pandemicAuthors: Jack K. Fitzsimons, Atul Mantri, Robert Pisarczyk, Tom Rainforth, Zhikuan ZhaoAbstract: …allowing additional precautions to be put in place to prevent continued transmission. Here we consider a cryptographic approach to contact tracing based on secure two-party computation (2PC). We begin by considering the problem of comparing a set of location histories held by two parties to determine whether they have come within some threshold distance whi… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05114 [pdf, other] quant-ph cond-mat.mes-hallNumber-resolved photocounter for propagating microwave modeAuthors: Rémy Dassonneville, Réouven Assouly, Théau Peronnin, Pierre Rouchon, Benjamin HuardAbstract: …. To maximize its performance, the device is first used as an \emph{in situ} waveform detector from which an optimal pump is computed and applied. Depending on the number of incoming photons, the detector succeeds with a probability that ranges from 56% to 99%. ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05102 [pdf, other] http://stat.ME stat.COA multi-resolution approximation via linear projection for large spatial datasetsAuthors: Toshihiro HiranoAbstract: …is subdivided, which leads to an approximated covariance function capturing both the large- and small-scale spatial variations. Moreover, we elicit the algorithms for fast computation of the log-likelihood function and predictive distribution with the approximated covariance function obtained by the M-RA-lp. Simulation studies and a real data analysis for… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 33 pages, 1 figure, 6 tablesarXiv:2004.05093 [pdf, ps, other] cond-mat.mes-hallOne-photon absorption by inorganic perovskite nanocrystals: A theoretical studyAuthors: T. P. T. Nguyen, S. A. Blundell, C. GuetAbstract: …theory. A confined exciton is described first within the Hartree-Fock (HF) approximation, and correlation between the electron and hole is then included in leading order by computing the first-order vertex correction to the electron-photon interaction. The vertex correction is found to give an enhancement of the near-threshold absorption cross section by a f… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 15 pages, 8 figuresarXiv:2004.05092 [pdf, other] http://math.AG math.ACToric varieties and Gröbner bases: the complete Q-factorial caseAuthors: Michele Rossi, Lea TerraciniAbstract: …associated toric varieties admit a projective embedding, in principle for any values of dimension and Picard number. The first algorithm is slower than the second one, but it computes all complete and simplicial fans supported by V and lead us to formulate a topological-combinatoric conjecture about the definition of a fan. On the other hand, we adapt th… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 18 pages, 2 figures. Final version accepted for publication in the special issue of the Journal AAAECC, concerning "Algebraic Geometry from an Algorithmic point of View"arXiv:2004.05084 [pdf] eess.IV cs.NEGSA-DenseNet121-COVID-19: a Hybrid Deep Learning Architecture for the Diagnosis of COVID-19 Disease based on Gravitational Search Optimization AlgorithmAuthors: Dalia Ezzat, Aboul ell Hassanien, Hassan Aboul EllaAbstract: In this paper, a novel approach called GSA-DenseNet121-COVID-19 based on a hybrid convolutional neural network (CNN) architecture is proposed using an optimization algorithm. The CNN architecture that was used is called DenseNet121 and the optimization algorithm that was used is called the gravitational search algorithm (GSA). The GSA is adapted to determine the best values for the hyperparameters… ▽ MoreSubmitted 9 April, 2020; originally announced April 2020.Comments: 29 pages - 7 Figures - 5 TablesarXiv:2004.05082 [pdf, other] eess.SP cs.LGAsynchronous Decentralized Learning of a Neural NetworkAuthors: Xinyue Liang, Alireza M. Javid, Mikael Skoglund, Saikat ChatterjeeAbstract: In this work, we exploit an asynchronous computing framework namely ARock to learn a deep neural network called self-size estimating feedforward neural network (SSFN) in a decentralized scenario. Using this algorithm namely asynchronous decentralized SSFN (dSSFN), we provide the centralized equivalent solution under certain technical assumptions. Asynchronou… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Journal ref: 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)arXiv:2004.05076 [pdf, other] cs.DB cs.NICheetah: Accelerating Database Queries with Switch PruningAuthors: Muhammad Tirmazi, Ran Ben Basat, Jiaqi Gao, Minlan YuAbstract: …and struggle to reduce query completion time with a large volume of data. In this paper, we leverage programmable switches in the network to partially offload query computation to the switch. While switches provide high performance, they have resource and programming constraints that make implementing diverse queries difficult. To fit in these constraints, w… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: To appear in ACM SIGMOD 2020arXiv:2004.05074 [pdf, ps, other] cs.DCPaxos vs Raft: Have we reached consensus on distributed consensus?Authors: Heidi Howard, Richard MortierAbstract: Distributed consensus is a fundamental primitive for constructing fault-tolerant, strongly-consistent distributed systems. Though many distributed consensus algorithms have been proposed, just two dominate production systems: Paxos, the traditional, famously subtle, algorithm; and Raft, a more recent algorithm positioned as a more understandable alternative to Paxos. In this paper, we consider t… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: To be published in the 7th Workshop on Principles and Practice of Consistency for Distributed Data (PaPoC)arXiv:2004.05071 [pdf, other] hep-th math-phFunctional Equations and Separation of Variables for Exact g-FunctionAuthors: Joao Caetano, Shota KomatsuAbstract: The g-function is a measure of degrees of freedom associated to a boundary of two-dimensional quantum field theories. In integrable theories, it can be computed exactly in a form of the Fredholm determinant, but it is often hard to evaluate numerically. In this paper, we derive functional equations---or equivalently integral equations of the thermodynamic Be… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 27 pages + appendicesarXiv:2004.05068 [pdf, other] cs.NEUncrowded Hypervolume-based Multi-objective Optimization with Gene-pool Optimal MixingAuthors: S. C. Maree, T. Alderliesten, P. A. N. BosmanAbstract: Domination-based multi-objective (MO) evolutionary algorithms (EAs) are today arguably the most frequently used type of MOEA. These methods however stagnate when the majority of the population becomes non-dominated, preventing convergence to the Pareto set. Hypervolume-based MO optimization has shown promising results to overcome this. Direct use of the hypervolume however results in no selection… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05065 [pdf, other] cs.DBOn Multiple Semantics for Declarative Database RepairsAuthors: Daniel Deutch, Amir Gilad, Sudeepa RoyAbstract: …alternative semantics, presenting the case for using each of them. We then study the relationships between the semantics in terms of their output and the complexity of computation. Our results formally establish the tradeoff between the permissiveness of the semantics and its computational complexity. We demonstrate th… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05050 [pdf] physics.flu-dynTheoretical Estimation of Attenuation Coefficient of Resonant Ultrasound Contrast AgentsAuthors: Lang XiaAbstract: …capable of estimating the attenuation coefficient due to the weakly nonlinear oscillations of the UCA were obtained with a relatively rigorous mathematical analysis. The computed results that were verified by numerical simulations showed the attenuation coefficient of the UCA at resonance was pressure-dependent and could be significantly smaller than that pr… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 24 pages (Accepted in JASA)arXiv:2004.05046 [pdf, other] cs.DCXChange: A Blockchain-based Mechanism for Generic Asset Trading In Resource-constrained EnvironmentsAuthors: Martijn de Vos, Can Umut Ileri, Johan PouwelseAbstract: …on multiple low-resource devices, we show that a full trade completes within half a second. To quantify the scalability of our mechanism, we conduct further experiments on our compute cluster. We conclude that the throughput of XChange, in terms of trades per second, scales linearly with the system load. Furthermore, we find that XChange exhibits superior th… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05038 [pdf, ps, other] math.CAAsymptotic computation of classical orthogonal polynomialsAuthors: A. Gil, J. Segura, N. M. TemmeAbstract: …are needed, the use of recursion to compute the polynomials is not a good strategy for… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: Contribution to EIBPOA2018arXiv:2004.05023 [pdf, other] stat.MERobust Estimation for Discrete-Time State Space ModelsAuthors: William H. Aeberhard, Eva Cantoni, Chris Field, Hans R. Kuensch, Joanna Mills Flemming, Ximing XuAbstract: …prove the Fisher consistency of our estimator and propose an implementation based on automatic differentiation and the Laplace approximation of integrals which yields fast computations. Simulation studies demonstrate that our robust procedure performs well both with and without deviations from model assumptions. Applying it to the stock assessment model for… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05022 [pdf, other] cs.DS cs.cc http://cs.DM math.COColouring (sP1+P5)-Free Graphs: a Mim-Width PerspectiveAuthors: Nick Brettell, Jake Horsfield, Daniel PaulusmaAbstract: …, and that there is a polynomial-time algorithm that, given a graph in the class, computes a branch decomposition of constant mim-width. A large number of \NP-complete graph problems become polynomial-time solvable on graph classes with bounded mim-width and for which a branch decomposition is quickly computable. The… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05020 [pdf, other] cs.CVModuleNet: Knowledge-inherited Neural Architecture SearchAuthors: Yaran Chen, Ruiyuan Gao, Fenggang Liu, Dongbin ZhaoAbstract: Although Neural Architecture Search (NAS) can bring improvement to deep models, they always neglect precious knowledge of existing models. The computation and time costing property in NAS also means that we should not start from scratch to search, but make every attempt to reuse the existing knowledge. In this paper, we discuss what kind of knowledge in… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05018 [pdf, other] cs.DS cs.cc http://cs.DM math.COBounding the Mim-Width of Hereditary Graph ClassesAuthors: Nick Brettell, Jake Horsfield, Andrea Munaro, Giacomo Paesani, Daniel PaulusmaAbstract: A large number of NP-hard graph problems become polynomial-time solvable on graph classes where the mim-width is bounded and quickly computable. Hence, when solving such problems on special graph classes, it is helpful to know whether the graph class under consideration has bounded mim-width. We first extend the toolkit for proving (un)boundedness of mim-wid… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.05016 [pdf, other] http://astro-ph.GA astro-ph.COCan a conditioning on stellar mass explain the mutual information between morphology and environment?Authors: Snehasish Bhattacharjee, Biswajit Pandey, Suman SarkarAbstract: …information between the two. Measuring the mutual information between morphology and environment by conditioning the stellar mass would allow us to test this possibility. We compute the mutual information between morphology and environment by conditioning the stellar mass in a volume limited and stellar mass limited sample from SDSS DR16 and find a non-zero… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 9 pages, 5 figures. Any comments and suggestions are welcomearXiv:2004.04999 [pdf, other] cs.SIEngagement Patterns of Peer-to-Peer Interactions on Mental Health PlatformsAuthors: Ashish Sharma, Monojit Choudhury, Tim Althoff, Amit SharmaAbstract: …a large-scale analysis of engagement patterns of 35 million posts on two popular online mental health platforms, TalkLife and Reddit. Leveraging communication models in human-computer interaction and communication theory, we operationalize a set of four engagement indicators based on attention and interaction. We then propose a generative model to jointly mo… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: Accepted to ICWSM 2020arXiv:2004.04995 [pdf, other] math.COThe 144 symmetries of the Littlewood-Richardson coefficients of SL3Authors: Emmanuel Briand, Mercedes RosasAbstract: We compute with SageMath the group of all linear symmetries for the Littlewood-Richardson associated to the representations of SL3. We find that there are 144 symmetries, more than the 12 symmetries known for the Littlewood-Richardson coefficients in general.Submitted 10 April, 2020; originally announced April 2020.Comments: 9 pages. See also the Sage Notebook as an ancillary file, or at: https://nbviewer.jupyter.org/url/emmanuel.jean.briand.free.fr/publications/the144symmetries.ipynbMSC Class: 05E10; 05E05arXiv:2004.04990 [pdf, ps, other] math.PRFluctuations of nodal sets on the 3-torus and general cancellation phenomenaAuthors: Massimo NotarnicolaAbstract: …number of nodal intersection points associated, respectively, with two and three independent arithmetic random waves of same frequency on the 3-torus. For these quantities, we compute their expected value, asymptotic variance as well as their limiting distribution. Our results are based on Wiener-Itô expansions for the volume and naturally complement the fin… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 59 pagesMSC Class: 60G60; 60B10; 60D05; 58J50; 35P20arXiv:2004.04986 [pdf, other] cs.LG http://cs.CR stat.MLTowards Realistic Byzantine-Robust Federated LearningAuthors: Amit Portnoy, Danny HendlerAbstract: Federated Learning (FL) is a distributed machine learning paradigm where data is decentralized among clients who collaboratively train a model in a computation process coordinated by a central server. By assigning a weight to each client based on the proportion of data instances it possesses, the rate of convergence to an accurate joint model can be greatly… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.ACM Class: I.2.6; I.2.11; C.2.0arXiv:2004.04985 [pdf, other] http://cs.IT cs.CRCoded Secure Multi-Party Computation for Massive Matrices with Adversarial NodesAuthors: Seyed Reza Hoseini Najarkolaei, Mohammad Ali Maddah-Ali, Mohammad Reza ArefAbstract: In this work, we consider the problem of secure multi-party computation (MPC), consisting of Γ sources, each has access to a large private matrix, N processing nodes or workers, and one data collector or master. The master is interested in the result of a polynomial function of the input matrices. Each source sends a randomized functions of its matrix, c… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 41 PagesarXiv:2004.04978 [pdf, ps, other] http://cs.NE cs.DS cs.LGA Simplified Run Time Analysis of the Univariate Marginal Distribution Algorithm on LeadingOnesAuthors: Benjamin Doerr, Martin KrejcaAbstract: With elementary means, we prove a stronger run time guarantee for the univariate marginal distribution algorithm (UMDA) optimizing the LeadingOnes benchmark function in the desirable regime with low genetic drift. If the population size is at least quasilinear, then, with high probability, the UMDA samples the optimum within a number of iterations that is linear in the problem size divided by the… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.04964 [pdf, other] astro-ph.HENeutrino physics in slowly rotating black hole spacetime and nonlinear electrodynamicsAuthors: H. Mosquera Cuesta, G. Lambiase, J. P. PereiraAbstract: …), as well as on the computation of that the electron fraction Ye, hence the r-processes, which may significantly differ with respect to the standard electrodynamics. ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 5 pages, 2 figures. Presented at NuPhys2019: Prospects in Neutrino Physics-Cavendish Conference Centre, London, 16-18 December 2019arXiv:2004.04961 [pdf, other] physics.comp-ph nucl-ex physics.med-phPreliminary results in using Deep Learning to emulate BLOB, a nuclear interaction modelAuthors: A. Ciardiello, M. Asai, B. Caccia, G. A. P. Cirrone, M. Colonna, A. Dotti, R. Faccini, S. Giagu, A. Messina, P. Napolitani, L. Pandola, D. H. Wright, C. Mancini-TerraccianoAbstract: …C reactions in the same energy domain. However, its computation time is too long for any medical application. For this reason we present the possibility of emulating it with a Deep Learning algorithm. Methods: The BLOB final state is a Probability Density Function (PDF) of finding a nucleon in a position of the phase space. We discretised this PDF and traine… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 8 pages, 9 figures, Accepted by Physica MedicaarXiv:2004.04948 [pdf, other] http://cs.IT cs.DC cs.LG eess.SPStraggler-aware Distributed Learning: Communication Computation Latency Trade-offAuthors: Emre Ozfatura, Sennur Ulukus, Deniz GunduzAbstract: When gradient descent (GD) is scaled to many parallel workers for large scale machine learning problems, its per-iteration computation time is limited by the straggling workers. Straggling workers can be tolerated by assigning redundant… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: This paper was presented in part at the 2019 IEEE International Symposium on Information Theory (ISIT) in Paris, France, and at the 2019 IEEE Data Science Workshop in Minneapolis, USAarXiv:2004.04938 [pdf, other] Inicio - Covarrubias & Cía cs.AIIdentifying Cultural Differences through Multi-Lingual WikipediaAuthors: Yufei Tian, Tuhin Chakrabarty, Fred Morstatter, Nanyun PengAbstract: …cross-cultural differences is an important application of natural language understanding. This problem is difficult due to the relativism between cultures. We present a computational approach to learn cultural models that encode the general opinions and values of cultures from multi-lingual Wikipedia. Specifically, we assume a language is a symbol of a cultu… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.04928 [pdf, ps, other] math.NAOptimal Subspace Expansion for Matrix Eigenvalue ProblemsAuthors: Zhongxiao JiaAbstract: …. By a careful analysis on the theoretical results, we obtain computable nearly optimal choices of wopt for the standard, harmonic and refined (harmonic) Rayleigh--Ritz methods. ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 15 pagesMSC Class: 65F15; 15A18; 65F10arXiv:2004.04912 [pdf, other] cs.CVPerson Re-Identification via Active Hard Sample MiningAuthors: Xin Xu, Lei Liu, Weifeng Liu, Meng Wang, Ruimin HuAbstract: …train a re-ID model from scratch. Then, intra-diversity estimation is designed to reduce the redundant hard samples by maximizing their diversity. Moreover, we propose a computer-assisted identity recommendation module embedded in the active hard sample mining framework to help human annotators to rapidly and accurately label the selected samples. Extensive… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.04910 [pdf, ps, other] http://cs.SC cs.ITA Simple Method for Computing Some Pseudo-Elliptic Integrals in Terms of Elementary FunctionsAuthors: Sam BlakeAbstract: We introduce a method for computing some pseudo-elliptic integrals in terms of elementary functions. The method is simple and fast in comparison to the algebraic case of the Risch-Trager-Bronstein algorithm. This method can quickly solve many pseudo-elliptic integrals which other well-known… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.arXiv:2004.04905 [pdf, other] http://math.CO math.LODistributed Algorithms, the Lovász Local Lemma, and Descriptive CombinatoricsAuthors: Anton BernshteynAbstract: …in the sense of topology or measure. To this end, we show that such well-behaved colorings can be produced using certain powerful techniques from finite combinatorics and computer science. First, we prove that efficient distributed coloring algorithms (on finite graphs) yield well-behaved colorings of Borel graphs of bounded degree; roughly speaking, determi… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 35 pagesarXiv:2004.04902 [pdf, other] Inicio - Covarrubias & Cía cs.LG cs.NEAn In-depth Walkthrough on Evolution of Neural Machine TranslationAuthors: Rohan Jagtap, Dr. Sudhir N. DhageAbstract: Neural Machine Translation (NMT) methodologies have burgeoned from using simple feed-forward architectures to the state of the art; viz. BERT model. The use cases of NMT models have been broadened from just language translations to conversational agents (chatbots), abstractive text summarization, image captioning, etc. which have proved to be a gem in their respective applications. This paper aims… ▽ MoreSubmitted 10 April, 2020; originally announced April 2020.Comments: 10 pages, 10 figuresFootnotes[1] Australians Accidentally Solve 58-Year-Old Quantum Mystery With Broken Lab Equipment[2] Quantum Computing Startup Raises $215 Million for Faster Device[3] ‘Amazing’ Math Bridge Extended Beyond Fermat’s Last Theorem