Transfer Learning for Human & AI

Multi-object tracking (MOT) enables mobile robots to perform well-informedmotion planning and navigation by localizing surrounding objects in 3D space and time . Existing methods rely on depth sensors (e.g., LiDAR) to detect andtrack targets in 3d space, but only up to a limited sensing range due to the sparsely-sensing range .…

This paper introduces a model of a stylized organization that is comprised of departments that autonomously allocate tasks . The organization guides the departments’ behavior by either an individualistic, a balanced, or an altruistic linear incentive scheme . Short-sighted decisions appear favorable since they do not only increase performance in the short run but also result in significantly higher performances in the long run .…

Stochastic gradient descent (SGD) is one of the most popular algorithms in machine learning . The noise encountered in these applications is different from that in theoretical analyses of stochastic gradientalgorithms . The energy landscape resembles that of overparametrized deep learning problems .…

Temporal Stream Logic (TSL) is a temporal logic that extends LTL with updatesand predicates over arbitrary function terms . This allows for specifying data-intensive systems for which LTL is not expressive enough . In TSL,functions and predicates are uninterpreted . We present an algorithm for checking the satisfiability of a TSL formula .…

This paper studies numerically how the signal detector arrangement influencesthe performance of reservoir computing using spin waves excited in aferrimagnetic garnet film . This investigation is essentially important sincethe input information is not only conveyed but also transformed by the spinwaves into high-dimensional information space when the waves propagate in the film in a spatially distributed manner .…

We propose here to make the connection between the definitions given by Turing and Wittgenstein about what it means to “follow a rule”. It will be here a presentation of the Turing test in order to observe that humans and machines have more in common than one might initially believe when it comes to interpreting signs .…

Collision avoidance algorithms are of central interest to many drone applications . In particular, decentralized approaches may be the key to enabling robust drone swarm solutions . By applying ageneral reinforcement learning approach, we propose a holistic learning-based approach to integrating collision avoidance with various tasks and dynamics .…

We propose here to make the connection between the definitions given by Turing and Wittgenstein about what it means to “follow a rule”. It will be here a presentation of the Turing test in order to observe that humans and machines have more in common than one might initially believe when it comes to interpreting signs .…

ICOS (Inlier searching usingCOmpatible Structures) is a novel, efficient and highly robust solution for the point cloud registration problem . ICOS is fast, accurate, robust against as many as 99% outliers with nearly 100% recall ratio of inlierswhether the scale is known or unknown, outperforming other state-of-the-art methods .…

Faces play a central role in the combinatorial and computational aspects of polyhedra . We present the first formalization of faces ofpolyhedra in the proof assistant Coq . We also prove a theorem due to Balinski on the$d$-connectedness of the adjacency graph of polytopes of dimension $d$ .…

Sentential Calculus with Identity (SCI) is an extension of classicalpropositional logic, featuring a new connective of identity between formulas . In SCI two formulas are said to be identical if they share the same denotation . In the semantics of the logic, truth values are distinguished from denotations,hence the identity connective is stronger than classical equivalence .…

The reversible computation paradigm aims to provide a new foundation for forgeneral classical digital computing that is capable of circumventing thethermodynamic limits to the energy efficiency of the conventional,non-reversible paradigm . We use the framework ofGorini-Kossakowski-Sudarshan-Lindblad dynamics (a.k.a. Lindbladians) with multiple asymptotic states, incorporating recent results from resource theory, full counting statistics, and stochastic thermodynamics .…

Distributed quantum systems and especially the Quantum Internet have the potential to fully demonstrate the power of quantumcomputing . Developing a general-purpose quantum computer is much more difficult than connecting many small quantum devices . One major challenge of implementing distributed quantum systems is programming them and verifying their correctness .…

We study multi-structural games, played on two sets of structures . These games generalize Ehrenfeucht-Fra\”{i}ss\'{e} games . We use these games to give a completecharacterization of the number of quantifiers required to distinguish linearorders of different sizes . We also develop machinery for analyzing structures beyond linear orders .…

Aggregates provide a concise way to express complex knowledge . Formalizing aggregates for answer set programming has proven to be challenging . We show that ternary satisfactionrelations bridge the gap between the standard Gelfond-Lifschitz reduct, and stable semantics as defined in the framework of ApproximationFixpoint Theory (AFT) We also show how different methods for handling aggregates taken from the literature can be described in terms of the framework and we study the corresponding Ternary Satisfaction Relations .…

When comparing learned image/video restoration and compression methods, it iscommon to report peak-signal to noise ratio (PSNR) results . However, there doesnot exist a generally agreed upon practice to compute PSNR for sets of imagesor video . Some authors report average of individual image/frame PSNR, which is equivalent to computing PSNR from geometric mean of individualimage/frame mean-square error (MSE) Others compute a single PSNR of Y-channel only, while others compute MSE/PSNR for RGB channels .…

Acting, stand-up and dancing are creative, embodied performances that follow a script . A slavishfollowing of the script leaves no room for creativity, but activeinterpretation of a script does . Theory and hypothesis are grounded in empirical research, and aim to provide resources for otherrobotic studies of the creative use of movement and gestures .…

The authors studied one such US Department of Energy mission-critical condensed matter physics application, Dynamical ClusterApproximation (DCA++) This paper discusses how device memory-bound challenges were successfully reduced by proposing an effective “all-to-all” communication method — a ring communication algorithm . The ring algorithm was optimized with sub-ring communicatorsand multi-threaded support to further reduce communication overhead and exposemore concurrency, respectively .…

We propose Super Odometry, a high-precision multi-modal sensor fusion framework . The proposed system was deployed on drones and groundrobots, as part of Team Explorer’s effort to the DARPA Subterranean Challengewhere the team won $1,000st and $2.2nd place in the Tunnel and UrbanCircuits, respectively.…

Reconfigurable intelligent surface (RIS) has emerged as a promising technology for achieving high spectrum and energy efficiency in future wireless networks . The RIS-aided single-cellmulti-user mobile edge computing (MEC) system is deployed to support the communication between a base station (BS) equipped with MEC servers and multiple single-antenna users .…

Glioblastoma Multiforme (GBM) is defined by an interplay of self renewal and invasiveness potential of multiple malignant glioma stem cells . The growth of GBM in a particular direction (node to stem) isrelated to an increase in mutual information . The findings are expected to contribute significantly in the design of future therapeutic mechanisms for GBM tumour .…

This article studies a novel distributed precoding design, coined teamminimum mean-square error (TMMSE) precoding . Building on the theory of teams, we derive a set of necessary and sufficientconditions for optimal TMMSE precoding, in the form of an infinite dimensionallinear system of equations .…

Array-like collection data structures are widely established in Python’s scientific computing-ecosystem for high-performance computations . High performance is,however, only guaranteed for static computations with a fixed computational domain . We show that for dynamic computations within an actively changing domain, the array-like collections provided by NumPy and itsderivatives are a bottleneck for large computations.…

The logics CS4 and IS4 are intuitionistic variants of the modal logic S4 . Whether the finite model property holds for each of these logics has been an open problem . In this paper we introduce two logics closelyrelated to IS4: GS4 and S4I, obtained by adding the Godel-Dummett axiom .…

The study investigates how popular the Nigerian TV reality show, Big Brother Naija BBN, is perceived by its viewers . Three hundred heavy viewers of the program were surveyed from Lagos and Ede, South-West Nigeria . Sixty six per cent 66 33.7 of respondents likethe program because it entertains.…

Reconfigurable intelligent surfaces (RIS) have been identified to be potential components of the future wireless networks . They can reconfigure thepropagation environment for wireless signals with low-cost passive devices . RIS has not only become anattractive research area but also triggered a couple of projects to develop solutions to enable the set-up of hardware demonstrations and prototypes .…

This is the first open-source memory compilers that has been developed specifically to automate ResistiveRandom Access Memory (RRAM) generation . RRAM holds the promise of achieving high speed, high density and non-volatility . A novel RRAM architecture,additionally is proposed, and a number of generated RRAM arrays are evaluated to identify their worst case control line parasitics and worst case settlingtime across the memristors of their cells .…

A parallel implementation of a compatible discretization scheme forsteady-state Stokes problems is presented in this work . The scheme usesgeneralized moving least squares to generate differential operators and applyboundary conditions . This meshless scheme allows a high-order convergence forboth the velocity and pressure, while also incorporates finite-difference-likesparse discretizing .…

This paper considers the use of two position receivers and an inertialmeasurement unit (IMU) to estimate the position, velocity, and attitude of arigid body, collectively called extended pose . Monte Carlo simulations demonstrate that the two-receiver IEKF approach yields improved estimates over the other filters considered .…

Hamming weights of sparse and long binary vectors are important modules in scientific applications, particularly in spiking neural networks that areof our interest . We propose a method inspired from synaptic transmissionfailure for exploiting FPGA lookup tables to compress long input vectors .…

Deep learning has solved difficult problems in visual patternrecognition, it is mostly successful in tasks where there are lots of labeled training data available . The brain is able to perform most of these tasks in a general way from limited to no labeled data .…

We propose here to make the connection between the definitions given by Turing and Wittgenstein about what it means to “follow a rule”. It will be here a presentation of the Turing test in order to observe that humans and machines have more in common than one might initially believe when it comes to interpreting signs .…

BioDynaMo is an open-source agent-based simulation platform that aims to alleviate researchers from the intricacies that go into the development of high-performance computing . Researchers in biology are faced with the tough challenge of developinghigh-performance computer simulations of their increasingly complex agents-basedmodels .…

Social networks represent nowadays in many contexts the main source of information transmission and the way opinions and actions are influenced . Modern marketing often involves paying some targeted users, or influencers, for advertising products or ideas . Find the set of nodes in a social network that lead to the highestinformation spread — the so-called Influence Maximization (IM) problem — is a pressing question and as such it has recently attracted a greatresearch interest .…

We present PSEUDo, an adaptive feature learning technique for exploring visual patterns in multi-track sequential data . We can even accomplish both themodeling and comparison of 10,000 different 64-track time series, each with 100time steps (a typical EEG dataset) under 0.8 seconds .…

In the last decade, deep neural networks have proven to be very powerful incomputer vision tasks . However, they are not robust toperturbations of the input data . Several methods able to generateadversarial samples make use of gradients, which usually are not available to an attacker in real-world scenarios .…

We study encodings of the lambdafail into the pi-calculus in the case of calculi with non-determinism and failures . Ourencoding precisely explains the interplay of non-Deterministic and fail-proneevaluation in lambfafail via typed processes in spi . In particular, it showshow failures in sequential evaluation (absence/excess of resources) can beneatly codified as interaction protocols .…

Deep Reinforcement Learning (DRL) is being increasingly applied to the problem of resource allocation for emerging System-on-Chip applications . The majority of SoC resource management approaches have been targeting makespanminimization with fixed number of jobs in the system . In contrast, SoCRATESaims at minimizing average latency in a steady-state condition while assigning tasks in the ready queue to heterogeneous resources .…

Predictive coding (PC) is a general theory of cortical function . The local, gradient-based learning rules found in one kind of PC model have recently been shown to closely approximate backpropagation . The model may also be useful for developing local learning algorithms that are compatible with neuromorphic hardware .…

Realistic simulators are critical for training and verifying roboticssystems . DriveGAN is a novel high-quality neural simulator called DriveGAN thatachieves controllability by disentangling different components withoutsupervision . It also includes controls for sampling features of a scene, such as the weather as well as the location of non-player objects .…

Deep Reinforcement Learning (DRL) is being increasingly applied to the problem of resource allocation for emerging System-on-Chip applications . In this paper, we introduce SoCRATES (SoCResource AdapTivE Scheduler), an extremely efficient DRL-based SoC scheduler . The resulting model is also compact insize and has very favorable energy consumption behaviors, making it highlypractical for deployment in future SoC systems with built-in neuralaccelerator.…

In discrete event simulation, event simultaneity occurs when any two events are scheduled to happen at the same point in simulated time . Simulation determinism is the expectation that the same semanticallyconfigured simulation will be guaranteed to repeatedly reproduce identical results .…

A bounded container maintains a collection of elements that can be inserted and extracted as long as the number of stored elements does not exceed the capacity . We consider the concurrent implementations of a bounded container more or less memory-friendly depending on how much memory they use in addition to storing the elements .…

Federated learning (FL) has recently emerged as an important and promising scheme in IoT, enabling devices to jointly learn a model without sharing their raw data sets . As the training data in FL is notcollected and stored centrally, FL training requires frequent model exchange, which is largely affected by the wireless communication network .…

We prove that (strong) fully-concurrent bisimilarity and causal-netbisimilarity are decidable for finite bounded Petri nets . The proofs are based on a generalization of the ordered marking proof technique used by Vogler .…

Permissionless blockchain protocols are known to consume an outrageous amount of computing power and suffer from a trade-off between latency and confidence in transaction confirmation . In this paper, we empower Algorand’s protocol by multiplexing itsbyzantine agreements in order to improve performance .…

In the scheduling with non-uniform communication delay problem, the input is a set of jobs with precedence constraints . Associated with every precedenceconstraint between a pair of jobs is a communication delay . The objective is to assign the jobs to machines to minimizethe makespan of the schedule .…

Modelling dynamics of urban venues is a challenging task as it ismultifaceted in nature . Demand is a function of many complex and nonlinearfeatures such as neighborhood composition, real-time events, and seasonality . Using a longitudinal dataset from Foursquare, we model individual venues and venuetypes across London and Paris .…