Fast uncertainty quantification of tracer distribution in the brain interstitial fluid with multilevel and quasi Monte Carlo

Efficient uncertainty quantification algorithms are key to understand thepropagation of uncertainty in high resolution mathematical models of brainphysiology . We show that QMC outperforms standard MCsimulations when the number of random inputs is small . MLMC considerablyoutperforms both QMC and standard MC methods and should therefore be preferredfor brain transport models, we say .…

Multichannel Analysis of Surface Waves Accelerated MASWAccelerated Software for Efficient Surface Wave Inversion Using MPI and GPUs

Multichannel Analysis of Surface Waves (MASW) is a technique frequently used in geotechnical engineering and engineering geophysics to infer layered modelsof seismic shear wave velocities in the top tens to hundreds of meters of thesubsurface . We aim to accelerate MASW calculations by capitalizing on modern computer hardware available in the workstations of most engineers: multiplecores and graphics processing units (GPUs) We propose new parallel and GPUaccelerated algorithms for evaluating MASW data, and provide softwareimplementations in C using Message Passing Interface (MPI) and CUDA .…

Identification of non local continua for lattice like materials

The paper focused on the dynamic homogenization of lattice-like materialswith lumped mass at the nodes to obtain energetically consistent models . The equation of motion of the lattice is transformed according to aunitary approach aimed to identify equivalent non-local continuum models ofintegral-differential and gradient type, the latter obtained through standard or enhanced continualization .…

Adaptive phase field modelling of crack propagation in orthotropic functionally graded materials

The proposed approach is capable of capturing the fracture process with a localized mesh refinement that provides notable gains in computational efficiency . The results reveal an increase in thestiffness and the maximum force with increasing material orientation angle . It is observedthat, if the gradation in fracture properties is neglected, the material gradient plays a secondary role, with the fracture behaviour being dominated by the orthotropy of the material.…

Bandits with adversarial scaling

We study a multi-armed bandit model where rewards have a stochastic and adversarial component . Our model captures display advertising where the “click-through-rate” can be decomposed to a (fixed acrosstime) arm quality component and a non-stochastic user-relevance component . We show that two algorithms, one from the action elimination and one fromthe mirror descent family are adaptive enough to be robust to adversarialscaling .…

Adaptation in Online Social Learning

This work studies social learning under non-stationary conditions . Classic social learning algorithms perform poorly under drifting conditions . We propose the Adaptive Social Learning (ASL) strategy . This strategy leverages an adaptiveBayesian update, where the adaptation degree can be modulated by tuning asuitable step-size parameter .…

A Distributed Pipeline for Scalable Deconflicted Formation Flying

A unified pipeline with onboard localization and adistributed, collision-free motion planning strategy that scales to a largenumber of vehicles . By enabling large-scale, deconflicted coordination, this pipeline should help pave the way for anytime, anywheredeployment of aerial swarms . The results show that our approach for solving the optimizationproblem associated with motion planning gives solutions within seconds in cases where general purpose solvers fail due to high complexity .…

Region adaptive graph fourier transform for 3d point clouds

The Region Adaptive Graph Fourier Transform (RA-GFT) is a multiresolutiontransform formed by combining spatially localized block transforms . We assume points are organized by a family of nested partitions represented by arooted tree . At each resolution level, attributes are processed in clustersusing block transforms.…

Language Integrated Updatable Views Extended version

Relational lenses are a modern approach to the view update problem in databases . They allow the definition of updatable views by the composition of lenses performing individual transformations . Horn et al. (2018) provided the firstimplementation of incremental relational lenses, which demonstrated thatrelational lenses can be implemented efficiently by propagating changes to the database rather than replacing the entire database state .…

Asynchronous effects

We explore asynchrony programming with algebraic effects . Decoupling execution of calls into signalling that an operation’s implementation needs to be executed . Interrupts a running computation with the operation’s result can react by installing interrupt handlers . We formalise these ideas in a small core calculus, called $\lambda_{\text{ae}}$ .…

Scaling MAP Elites to Deep Neuroevolution

Quality-Diversity (QD) algorithms, and MAP-Elites (ME) in particular, have proven very useful for a range of applications including enabling real robots to recover quickly from joint damage, solving strongly deceptive mazetasks or evolving robot morphologies to discover new gaits . We propose to leverage the efficiency of Evolution Strategies (ES) to scale MAP-elites to high-dimensional controllers with large neural networks .…

Robust Market Making via Adversarial Reinforcement Learning

We show that adversarial reinforcement learning (ARL) can be used to produce market marking agents that are robust to adversarial and adaptively-chosen conditions . We turn the well-studied single-agent model of a zero-sum game between amarket maker and adversary . The adversary acts as a proxy for other market participants that would like to profit at the market maker’s expense .…

GenNet Reading Comprehension with Multiple Choice Questions using Generation and Selection model

Multiple-choice machine reading comprehension is difficult task as its required machines to select the correct option from a set of candidate orpossible options using the given passage and question . Here we proposed GenNet model, a neural network-basedmodel . In this model first we will generate the answer of the question from thepassage and then will matched the generated answer with given answer, the bestmatched option will be our answer .…

Hierarchically Decoupled Imitation for Morphological Transfer

Learning long-range behaviors on complex high-dimensional agents is afundamental problem in robot learning . For such tasks, we argue that transferring learned information from a morphologically simpler agent can improve the sample efficiency of a more complex one . We propose a hierarchical decoupling of policies into two parts: an independentlylearned low-level policy and a transferable high level policy .…

Model Assertions for Monitoring and Improving ML Models

ML models are increasingly deployed in settings with real world interactions such as vehicles . We propose a new abstraction, model assertions, that adapts theclassical use of program assertions as a way to monitor and improve ML models . For training, we propose a bandit-basedactive learning algorithm that can sample from data flagged by assertions .…

Embodied Synaptic Plasticity with Online Reinforcement learning

The endeavor to understand the brain involves multiple collaborating researchfields . We demonstrate this framework to evaluate Synaptic Plasticitywith Online REinforcement learning (SPORE), a reward-learning rule based onsynaptic sampling, on two visuomotor tasks . The resulting framework allows to evaluatethe validity of biologically-plausibe plasticity models in closed-loop roboticsenvironments .…

Relevance Guided Modeling of Object Dynamics for Reinforcement Learning

Current deep reinforcement learning approaches incorporate minimal priorknowledge about the environment, limiting computational and sample efficiency . In this paper we propose aframework for reasoning about object dynamics and behavior to rapidly determineminimal and task-specific object representations . We also demonstrate the potential of this framework on a number of Atari games, using our object representation and standard RL and planningalgorithms to learn over 10,000x faster than standard deep RL algorithms .…

Learning Context aware Task Reasoning for Efficient Meta reinforcement Learning

Despite recent success of deep network-based Reinforcement Learning (RL) it remains elusive to achieve human-level efficiency in learning novel tasks . We decompose the meta-RL problem into three sub-tasks, task-exploration,task-inference and task-fulfillment . We validate our approach with extensive experiments on several benchmarks and the results show that our algorithm effectively performsexploration for task inference, improves sample efficiency during both training and testing, and mitigates the Meta-overfitting problem .…

Digital Collaborator Augmenting Task Abstraction in Visualization Design with Artificial Intelligence

In the task abstraction phase of the visualization design process, a practitioner maps the observed domain goals to generalizable abstract tasks using visualization theory in order to better understand and address the users needs . We argue that this manual taskabstraction process is prone to errors due to designer biases and a lack of domain background and knowledge .…

Two Decades of AI4NETS AI ML for Data Networks Challenges Research Directions

The popularity of Artificial Intelligence (AI) has dramatically increased in the last few years . Despite the many attempts to turn networks into learning agents, the successful application of AI/ML in networking is limited . There are stillmany unsolved complex challenges associated to the analysis of networking datathrough AI/ml, which hinders its acceptability and adoption in the practice .…

Direct Product Primality Testing of Graphs is GI hard

In [1] Imrich proves that both primality testing and aunique prime factorization can be determined in polynomial time for (finite)connected and nonbipartite graphs . The author states as an open problem how results on the direct product of nonbipsartite, connected graphs extend tobipartites connected graphs and to disconnected ones .…

Sparse Tiling through Overlap Closures for Termination of String Rewriting

A sparse set of tiles contains only those that are reachable in derivations, and is constructed by completing an automaton . We over-approximate reachability sets in string rewriting by languages defined by admissible factors, called tiles . Using a partial algebra defined by a sparse tiling for semanticlabelling, we obtain a transformational method for proving local termination .…

A Cubical Language for Bishop Sets

XTT is a version of Cartesian cubical type theory specialized for Bishop sets . Every type enjoys a definitional version of the uniqueness of identity proofs . XTT reconstructs ideas underlying Observational Type Theory . We prove thecanonicity property of XTT (that every closed boolean is definitionally equal to a constant) by Artin gluing .…

The Prolog Debugger and Declarative Programming Examples

This paper contains examples for a companion paper “The Prolog Debugger andDeclarative Programming” The companion paper triesto find methods of using it from the declarative point of view . Prolog makes logic programming possible, at least to a substantial extent, but the Prolog debugger works solely in terms of the operational semantics .…

Anchor Attention for Hybrid Crowd Forecasts Aggregation

State-of-the-art forecasting platforms are “hybridized” They gather forecasts from a crowd of humans, as well as one or more machine models . We propose anchor attention for this type of sequence summary problem . We evaluate our approach using data from real-world forecastingtournaments, and show that our method outperforms the current state-of theartaggregation approaches .…

Bringing Inter Thread Cache Benefits to Federated Scheduling Extended Results Technical Report

Multiprocessor scheduling of hard real-time tasks modeled by directed acyclicgraphs (DAGs) exploits the inherent parallelism presented by the model . The DAG-OT model with cache-aware scheduling reduces the numberof cores allocated to individual tasks by approximately 20 percent in the synthetic evaluation and up to 50 percent on a novel parallel computingplatform implementation .…

Sealing Pointer Based Optimizations Behind Pure Functions

Functional programming languages are particularly well-suited for buildingautomated reasoning systems . But existing languages suffer a major limitation in these domains: traversing a term requires time proportional to the tree size of the term as opposed to its graphsize . We show how to use dependent types to seal the necessary pointer-address manipulations behind pure functional interfaces .…

Data Migration using Datalog Program Synthesis

This paper presents a new technique for migrating data between differentschemas . Our method expresses the schemas mapping as a Datalog program andautomatically synthesizes a program from simple input-output examplesto perform data migration . We implement the proposed technique as a tool called Dynamite and show itseffectiveness by evaluating Dynamite on 28 realistic data migration scenarios .…

Robot Mindreading and the Problem of Trust

This paper raises three questions regarding the attribution of beliefs,desires, and intentions to robots . The first one is whether humans in factengage in robot mindreading, if they do, this raises a second question: doesrobot mindreading foster trust towards robots? The last part of the paper discusses different ways toanalyze this apparent trade-off and suggests that a possible solution mustadopt tolerable degrees of opacity that depend on pragmatic factors connected to the level of trust required for the intended uses of the robot .…

Natural Language Processing Advancements By Deep Learning A Survey

Natural Language Processing (NLP) helps empower intelligent machines by enhancing a better understanding of the human language for linguistic-basedhuman-computer communication . Survey categorizes and addresses the different aspects and applications of NLP that have benefited from deep learning . It covers core NLP tasks and applications and describes how deep learning methods and models advance theseareas.…

Real World Human Robot Collaborative Reinforcement Learning

The intuitive collaboration of humans and intelligent robots is an essential objective for many desirable applications of robots . We present a real-world setup of a human-robot collaborative maze game, designed to be non-trivial and only solvable through collaboration . We use deep reinforcement learning for the control of the robotic agent, and achieve results within 30 minutes of play, without any type of pre-training .…

ProxEmo Gait based Emotion Learning and Multi view Proxemic Fusion for Socially Aware Robot Navigation

ProxEmo is a novel end-to-end emotion prediction algorithm forsocially aware robot navigation among pedestrians . Our approach predicts theperceived emotions of a pedestrian from walking gaits, which is then used foremotion-guided navigation taking into account social and proxemic constraints . It achieves a meanaverage emotion prediction precision of 82.47% on the Emotion-Gait benchmarkdataset .…

D3VO Deep Depth Deep Pose and Deep Uncertainty for Monocular Visual Odometry

We propose D3VO as a novel framework for monocular visual odometry thatexploits deep networks on three levels — deep depth, pose and uncertaintyestimation . The proposed network outperforms state-of-the-art monocular VO methods by a largemargin . It also achieves comparable results to stereo/LiDARodometry on KITTI and to the state of the art visual-inertial odometry on EuRoCMAV, while using only a single camera .…

Learning and Solving Regular Decision Processes

Regular Decision Processes (RDPs) are a recently introduced model that extends MDPs with non-Markovian dynamics and rewards . Learning RDPs from data is a challenging problem that hasyet to be addressed, on which we focus in this paper . Our approach rests on anew representation using Mealy Machines that emit a distribution and an expected reward for each state-action pair.…