COMPOSE Cross Modal Pseudo Siamese Network for Patient Trial Matching

Clinical trials play important roles in drug development but often suffer from expensive, inaccurate and insufficient patient recruitment . Theavailability of massive electronic health records (EHR) data and trialeligibility criteria (EC) bring a new opportunity to data driven patientrecruitment . CrOss-Modal PseudO-SiamEsenetwork (COMCOMpose) proposed CrOSS-modal PseudoO-PseudO to address these challenges for patient-trial matching .…

Pessimism About Unknown Unknowns Inspires Conservatism

If we could define the set of all bad outcomes, we could hard-code an agent which avoids them . We do not know of any general-purpose approaches in the literature to avoiding novel failure modes . Motivated by this, we define an idealizedBayesian reinforcement learner which follows a policy that maximizes the worst-case expected reward over a set of world-models .…

Certifying Strategyproof Auction Networks

Optimal auctions maximize a seller’s expected revenue subject to individualrationality and strategyproofness for the buyers . Myerson’s seminal work in 1981 settled the case of auctioning a single item . Recent thread of work in “differentiable economics” has used tools from modern deep learning to instead learn good mechanisms .…

Latent Bandits Revisited

A latent bandit problem is one in which the learning agent knows the armreward distributions conditioned on an unknown discrete latent state . The goal of the agent is to identify the latent state, after which it canact optimally . This setting is a natural midpoint between online and offlinelearning .…

Causal intersectionality for fair ranking

Rankings are used in many contexts, ranging from Web searchresults to college admissions, but causal inference for fair rankings hasreceived limited attention . In this paper we propose a causal modeling approach to intersectionalfairness, and a flexible, task-specific method for computing intersectionallyfair rankings .…

Pipeline PSRO A Scalable Approach for Finding Approximate Nash Equilibria in Large Games

Policy SpaceResponse Oracles (PSRO) is a deep reinforcement learning algorithm that is guaranteed to converge to an approximate Nash equilibrium . However, PSRO requires training a reinforcement learning policy at eachiteration, making it too slow for large games . Pipeline PSRO(P2SRO) is the first scalable general method for finding approximate Nashequilibria in large zero-sum imperfect-information games .…

Efficient Reasoning in Regular Boardgames

Regular Boardgames (RBG) language is a universal General Game Playing formalism for the class offinite deterministic games with perfect information . The impact of these optimizations ranges from 1.7 to even 33-fold efficiency improvement when measuring the number of possible game playouts persecond .…

p d Separation A Concept for Expressing Dependence Independence Relations in Causal Networks

Spirtes, Glymour and Scheines formulated a Conjecture that a direct dependence test and a head-to-head meeting test would suffice to construedirected acyclic graph decompositions of a joint probability distribution(Bayesian network) for which Pearl’s d-separation applies . This paper uses the concept of p-d separation (partial dependency separation) to prove that this is true .…

Piecewise Stationary Off Policy Optimization

Off-policy learning is a framework for evaluating and optimizing policies without deploying them, from data collected by another policy . Real-world environments are typically non-stationary and the offline learned policiesshould adapt to these changes . This approach ispractical and analyzable, and we provide guarantees on both the quality ofoff-policy optimization and the regret during online deployment.…

Learn to Effectively Explore in Context Based Meta RL

Meta reinforcement learning (meta-RL) provides a principled approach for fastadaptation to novel tasks by extracting prior knowledge from previous tasks . The explorer is motivated by an information-theoretical intrinsic reward that encourages the explorer to collect experiences that provide rich information about the task .…

Selection of an Integrated Security Area for locating a State Military Organization SMO based on group decision system a multicriteria approach

High crime rates in Brazil have encouraged authorities to identify solutions to minimize crimes . Each ISA has a neighborhood conglomerates taking into account their location . From this it becomes possible to maximize security management and combat crime . The paper aims to identify the best ISA to deploy a police battalion using Group Decision SupportSystem (GDSS) Called GRoUp Support (GRUS) was used from two main Votetechniques: Condorcet and Borda.…

A VIKOR and TOPSIS focused reanalysis of the MADM methods based on logarithmic normalization

Logarithmic Normalization(LN) method has a distinguished advantage, reflecting in that a sum of the normalized values of criteria always equals 1 . VIKOR and TOPSIS, as the two famous MADM methods, were selected for this reanalysis research study . The results indicate that there are differences between the two approaches to the two MADM approaches, which are based on different MADM techniques, according to the research .…

Improved algorithm for permutation testing

We study the problem of testing forbidden patterns . The patterns that are ofsignificant interest include monotone pattern and $(1,3,2)$-pattern . We provide a simple adaptivealgorithm with one-sided error for testing monotones . We alsopresent an algorithm with improved query complexity for testing$(1, 3,2), $O(O(k^2) and $O(\log n)…

Algorithmically Optimal Outer Measures

Maintheorem states that the classical local fractal dimensions of any locallyoptimal outer measure coincide exactly with the algorithmic dimensions of outer measures in Euclidean spaces . We discussimplications for point-to-set principles . We introduce global and local optimality conditions for lower semicomputableouter measures .…

Existential Theory of the Reals Completeness of Stationary Nash Equilibria in Perfect Information Stochastic Games

We show that the problem of deciding whether in a multi-player perfectinformation recursive game (i.e. a stochastic game with terminal rewards) thereexists a stationary Nash equilibrium ensuring each player a certain payoff is Existential Theory of the Reals complete . Our result holds for acyclic games,where a Nash equilibrium may be computed efficiently by backward induction .…

The PSPACE hardness of understanding neural circuits

In neuroscience, an important aspect of understanding the function of aneural circuit is to determine which, if any, of the neurons in the circuit are vital for the biological behavior governed by the neural circuit . Recent advances in experimental techniques have provided researchers with tools to activate and deactivate subsets of neurons with a very highresolution, even in living animals .…

Solving the Bethe Salpeter equation on massively parallel architectures

The last ten years have witnessed fast spreading of massively parallelcomputing clusters . The seamless integration ofsoftware and middleware libraries is a key ingredient to ensure portability of scientific codes . In this work, wedescribe the integration of the ChASE library, a modern parallel eigensolver, into an existing legacy code for the first-principles computation of opticalproperties of materials .…

Geo PIFu Geometry and Pixel Aligned Implicit Functions for Single view Human Reconstruction

Geo-PIFu is a method to recover a 3D mesh from a monocular colorimage of a clothed person . Our method is based on a deep implicitfunction-based representation to learn latent voxel features using astructure-aware 3D U-Net . We show that, by both encoding query points and constraining globalshape, the reconstruction we obtain for clothedhuman meshes exhibits less shape distortion and improved surface details .…

ShapeFlow Learnable Deformations Among 3D Shapes

ShapeFlow is a flow-based model for learning a deformation space forentire classes of 3D shapes with large intra-class variations . ShapeFlow allowslearning a multi-template deformation . space that is agnostic to shape topology,yet preserves fine geometric details . We parametrizethe deformation between geometries as a learned continuous flow field via aneural network .…

Repulsive Curves

Most tools for curve design do nothing toprevent crossings or self-intersections . This paper develops efficient algorithms for (self-)repulsion of plane and space curves that are well-suited to problems in computational design . The energy is easily integrated with a variety of constraints and penalties (e.g.,…

Sound Search in Imperfect Information Games

Search has played a fundamental role in computer game research since the verybeginning . While online search has been commonly used in perfectinformation games such as Chess and Go, online search methods for imperfectinformation games have only been introduced relatively recently .…

Blocking defector invasion by focusing on the most successful partner

According to the standard protocol of spatial public goods game, a cooperator player invests not only into his own game but also into the games organized by his neighbors . Weshow that this very simple alteration of the dynamical rule results in asurprisingly positive evolutionary outcome — cooperators prevail even in harshenvironment represented by small values of the synergy factor in the game .…

User Profiling from Reviews for Accurate Time Based Recommendations

Product reviews on sites such as Amazon represent a valuable datasource to understand why someone bought an item and potentially who the item isfor . This information can then be used to construct a dynamic user profile . Mining temporally related content in reviews can enable the recommenderto go beyond finding similar items or users to potentially predict a futureneed of a user .…