A non cooperative meta modeling game for automated third party calibrating validating and falsifying constitutive laws with parallelized adversarial attacks

The evaluation of constitutive models, especially for high-risk and high-regret engineering applications, requires efficient and rigorousthird-party calibration, validation and falsification . We introduce an automated meta-modelinggame where two competing AI agents systematically generate experimental data tocalibrate a given constitutive model and to explore its weakness .…

Adversarial Augmentation Policy Search for Domain and Cross Lingual Generalization in Reading Comprehension

Reading comprehension models often overfit to nuances of training datasetsand fail at adversarial evaluation . Training with adversarially augmenteddataset improves robustness against those adversarial attacks but hurts generalization of the models . We propose three new methods for generating QAadversaries, that introduce multiple points of confusion within the context, show dependence on insertion location of the distractor, and reveal the compound effect of mixing adversarial strategies with syntactic and semanticparaphrasing methods .…

BLEU might be Guilty but References are not Innocent

The quality of automatic metrics for machine translation has been called into question, especially for high-quality systems . Thispaper demonstrates that, while choice of metric is important, the nature of thereferences is also critical . Ourmethod yields higher correlation with human judgment not only for thesubmissions of WMT 2019 English to German, but also for Back-translation andAPE augmented MT output, which have been shown to have low correlation with automatic metrics using standard references .…

Technical Report NEMO DNN Quantization for Deployment Model

This technical report aims at defining a formal framework for Deep NeuralNetwork (DNN) layer-wise quantization . It also acts as a documentation for the NEMO(NEural Minimization for pytOrch) framework . An important feature of this model is that it enables DNN inference using purelyintegers – without resorting to real-valued numbers in any part of thecomputation and without relying on an explicit fixed-point numericalrepresentation .…

Quantifying Notes Revisited

In this survey we review several dynamic epistemic logics with modalitiesrepresenting quantification over information change . Of such logics we presenttypical axioms involving the relation between knowledge or belief and belief andinformative action . We focus on open problems and new directions for research .…

Generating Fact Checking Explanations

Most existing work on automated fact checking is concerned with predicting veracity of claims based on metadata, social network spread, language usedin claims, and evidence supporting or denying claims . A crucial piece of the puzzle that is still missing is to understand how to automate themost elaborate part of the process — generating justifications for verdicts on claims .…

Exact and Approximate Algorithms for Computing a Second Hamiltonian Cycle

Cedric Smith proved in 1946, using a non-constructive parity argument, that such a second Hamiltonian cycle always exists . Ouralgorithm is based on a fundamental structural property of Thomason’s lollipopalgorithm, which we prove here for the first time . This approximation result also improves the state of the art running time for solving this problem, as well as improving the state-of-the-art running time of the algorithm for solving such a problem .…

Optimizing Reachability Sets in Temporal Graphs by Delaying

A temporal graph is a dynamic graph where every edge is assigned a set of integer time labels that indicate at which discrete time step the edge is available . In this paper, we study how changes of the time labels,corresponding to delays on the availability of the edges, affect thereachability sets from given sources .…

Embedded model discrepancy A case study of Zika modeling

Mathematical models of epidemiological systems enable investigation of and predictions about potential disease outbreaks . However, commonly used models are often highly simplified representations of incredibly complex systems . We propose an embedded discrepancy operator, a modification to the model equations that requires modest information about the system and is calibrated by all relevant data .…

Complaint driven Training Data Debugging for Query 2 0

Rain is a complaint-driven training data debuggingsystem . Rain allows users to specify complaints over the query’s intermediateor final output, and aims to return a minimum set of training examples . Rain achieves the highest recall@k among all the baselines while still returns results interactively .…

Workflow Automation for Cyber Physical System Development Processes

Development of Cyber Physical Systems (CPSs) requires close interaction between developers with expertise in many domains to achieve ever-increasing demands for improved performance, reduced cost, and more system autonomy . Eachengineering discipline commonly relies on domain-specific modeling languages . However, integration between these heterogeneous models and tools is lacking .…

From Holant to Quantum Entanglement and Back

Holant problems are intimately connected with quantum theory as tensornetworks . We use techniques from Holant theory to derive new and improved results for quantum entanglement theory . The signatures need not be symmetric, and noauxiliary signatures are assumed . We discover two particular entangledstates $|{\Psi_6}\rangle$ of 6 qubits and $|\/\/\/8) that have extraordinary and unique closure properties in terms of the Bell property .…

SESAME Semantic Editing of Scenes by Adding Manipulating or Erasing Objects

SESAME is a novel generator-discriminator pair for Semantic Editing ofScenes by Adding, Manipulating or Erasing objects . In our setup, the userprovides the semantic labels of the areas to be edited and the generatorsynthesizes the corresponding pixels . We evaluate our model on a diverse set ofdatasets and report state-of-the-art performance on two tasks: (a) imagemanipulation and (b) image generation conditioned on semantic labels .…

Closing Gaps in Asymptotic Fair Division

We study a resource allocation setting where $m$ discrete items are to bedivided among $n$ agents with additive utilities . We prove that the classical round-robinalgorithm is likely to produce an envy-free allocation . We show that a proportional allocation exists with high probability as longas $m\geq n$, while an allocation satisfying envy-freeness up to any item (EFX)is likely to be present for any relation between $m and $n$.…

Image Co skeletonization via Co segmentation

Image co-skeletonization is defined as jointskeleton extraction of objects in an image collection . Object skeletonization in a single natural image is a challenging problem because there is hardly anyprior knowledge about the object . We hope that the commonness prior that exists across theimages may help, just as it does for other joint processing problems such asco-segmentation .…

μ Tiles Efficient Intra Process Privilege Enforcement of Memory Regions

The status quo is that Unix-like operating systems do not offer privilegeseparation inside a process . Lack of practical fine-grainedcompartmentalization inside a shared address space leads to private dataleakage through applications’ untrusted dependencies and compromised threads . To this end, we propose a lightweight kernel abstraction and set of security primitives based on mutual distrust .…

Norm violation in online communities A study of Stack Overflow comments

Online communities are also expected to abide by the rules and regulations that are expressed in the code of conduct of a system . Hate speech and abusive language usage are on the rise among StackOverflow users . By identifying andminimising various patterns of norm violations in different SO communities, the community would become less toxic and thereby the community can engage more effectively in its goal of knowledge sharing .…

Probabilistic Resource Aware Session Types

Session types guarantee that message-passing processes adhere to predefined protocols . Probabilistic session types describe probability distributions over messages and are aconservative extension of intuitionistic (binary) session types . The analysis for expected resource bounds is integratedwith the type system and is a variant of automatic amortized resource analysis.…

Computing with Continuous Objects A Uniform Co inductive Approach

A uniform approach to computing with infinite objects like real numbers,tuples of these, compacts sets, and uniformly continuous maps is presented . The approach is of the same power as Weihrauch’s Type-Two Theory of Effectivity . It is based on Berger and the present author generalised this approach to complete metric spaces and showed how to deal with compact sets .…

Social rules for agent systems

When creating (open) agent systems it has become common practice to uses social concepts such as social practices, norms and conventions to model the interactions between the agents are regulated . In this paper we will investigate some of the relations between these concepts and also whether they are fundamentally stemming from a single social object or should be seen as different types of objects altogether .…

Application of Just Noticeable Difference in Quality as Environment Suitability Test for Crowdsourcing Speech Quality Assessment Task

Crowdsourcing micro-task platforms facilitate subjective media qualityassessment by providing access to a highly scale-able, geographicallydistributed and demographically diverse pool of crowd workers . Environmental noise at the listener’s side and the listening device (loudspeaker or headphone) significantly affect perceived quality, and consequently the reliability and validity of subjective ratings .…

Increasing Validity Through Replication An Illustrative TDD Case

Software Engineering (SE) experiments suffer from threats tovalidity that may impact their results . Replication allows researchers building on top of previous experiments’ weaknesses and increasing the reliability of the findings . Differences across TDD replications’ results might be due to the operationalization of the response variables, the allocation of subjects totreatments, the allowance to work outside the laboratory, the provision of the task, or the task .…

A Procedure and Guidelines for Analyzing Groups of Software Engineering Replications

Researchers from different groups and institutions are collaborating on building groups of experiments by means of replication (i.e., conductinggroups of replications) Disparate aggregation techniques are being applied toanalyze groups of replication . The application of unsuitable techniques toaggregate replication results may undermine the potential of groups ofreplications to provide in-depth insights from experiment results .…