Transfer Learning for Human & AI

What does the age of participation look like from the perspective of a museum visitor? Arguably, the concept of participative experiences is already sodeeply ingrained in our culture that we may not even think about it as participation . How can museums tap into these activities – and make themselves relevant to visitors?…

Gaussian process regression is used to learn data sets collected from sensors in Internet-of-Things systems . Data-aided sensing is generalized for distributedselective uploading when sensors can have feedback of predictions of their measurements so that each sensor can decide whether or not it uploads by comparing its measurement with the predicted one .…

This paper presents a review of the design and application of modelpredictive control strategies for Micro Aerial Vehicles and specifically multirotor configurations such as quadrotors . The diverse set of works in the domain is organized based on the control law being optimized over linear or nonlinear dynamics, the integration of state and input constraints, possiblefault-tolerant design, if reinforcement learning methods have been utilized and if the controller refers to free-flight or other tasks such as physicalinteraction or load transportation .…

We propose a novel numerical approach to separate multiple tissuecompartments in image voxels . We estimate quantitatively their nuclearmagnetic resonance (NMR) properties and mixture fractions, given magneticresonance fingerprinting (MRF) measurements . The number of tissues, their types or quantitative properties are not a-priori known, but the image is assumed to be composed of sparse compartments with linearly mixed Bloch magnetisationresponses .…

In this paper we revisit one of the classical statistical problems, theso-called sparse maximum-likelihood (ML) linear regression . As a way ofattacking this type of regression, we present a novel CLuP mechanism that relies on the Random Duality Theory (RDT) basedalgorithmic machinery that we recently introduced in  Stojnicclupint19, StojnicClupcmpl19,StojNicclupplt19, and Stojnicluprephased20 .…

A ‘one-size-fits-all’ approach is often sub-optimal as the bestvalue depends on runtime context . We present Resonance, asystem based on contextual bandits (CB) We describe experiences from threereal-world experiments: applying it to the audio, video, and transport components in Skype .…

This paper is devoted to the investigation of inverse problems related to drift-diffusion equations modeling semiconductor devices . In thiscontext we analyze several identification problems corresponding to different types of measurements, where the parameter to be reconstructed is aninhomogeneity in the PDE model (doping profile) For a particular type ofmeasurement (related to the voltage-current map) we consider special cases of Drift-Diffusion equations .…

This paper studies the Two-Person Zero Sum(TPZS) game between aMultiple-Input Multiple-Output(MIMO) radar and an extended target . payofffunction is the output Signal-to-Interference-pulse-Noise Ratio(SINR) at theradar receiver . The Stackelberg equilibriumstrategy of radar, namely robust or minimax waveform-filter pair, for threedifferent cases are taken into consideration .…

Many of today’s robot perception systems aim at accomplishing perception tasks that are too simplistic and too hard . We propose a perception system that maintains its beliefs about its environment as a scene graph with physics simulation and visual rendering .…

In this paper, we present and analyze a weak Galerkin finite element (WG) method for solving the symmetric hyperbolic systems . This method is highlyflexible by allowing the use of discontinuous finite elements on element and boundary independently of each other .…

Family of conservative one-stepschemes of the sixth order of accuracy for Zakharov-Shabat system is constructed . The exponential integrator is a special case . Schemes based onrational approximation allow the use of fast algorithms to solve the initialproblem for a large number of values of the spectral parameter .…

Sylvester, Lyapunov, and algebraic Riccati equations are the bread and butter of control theorists . They are used to compute infinite-horizon Gramians, solve optimal control problems in continuous or discrete time, and design observers . Here, we derive the forward and reverse-mode derivatives of the solutions to all three types of equations, and showcase their application on an inverse control problem .…

Decimal multiplication is the task of multiplying two numbers in base $10^N.$ Using only portable techniques, we achieve a 3x—5x speedup over the mpdecimal library . We also present a simple cache-efficient algorithm for in-place $2n \times n$ or $n\times 2n$ matrix transposition .…

In order to achieve real-time operation, existing approaches often assume previously-estimated states to be perfectly known . The Schmidt-Kalman filter has processing cost linear in the size of the state vector but quadratic memory requirements . Inparticular, this method, the resource-aware inverse Schmidt estimator (RISE)allows estimation accuracy for computational efficiency .…

The proposed medium access control (MAC) design is designed for machine-type communications in the industrial internet of things . The proposed MAC can support 1000 devices under an aggregated traffic load of 3000 packets per second with a single channel and achieve <0.5ms average delay and <1% average collision probability among 50high priority devices . The two-stepapproach ensures the accuracy and granularity necessary for satisfying the QoSrequirements and avoids excessive complexity from handling a large number of devices. Integrating the distributed coordination in the protocol design fromPart I and the centralized scheduling from this part, the proposed MAC protocolachieves high performance, demonstrated through extensive simulations. Forexample, the results show that the proposedMAC can support1000 devices underan aggregated Traffic load of 3,000 packets perSecond with a one channel andachieve < 0.5 ms average delay, respectively . The proposal is demonstrated through simulations. The proposedMAC protocolachieve high performance. The proposal was demonstrated through . …

SARL-SGAN-KCE combines a deep socially aware attentive valuenetwork with a human multimodal trajectory prediction model . Thekinematic constraints of the vehicle are also considered to ensure smooth and safe trajectories . We evaluate our method against the state-art methods forcrowd navigation and provide an ablation study to show that our method is safer and closer to human behaviour .…

Deformable objects present a formidable challenge for robotic manipulation due to the lack of canonical low-dimensional representations and the difficulty of capturing, predicting, and controlling such objects . We construct compacttopological representations to capture the state of highly deformable objectsthat are topologically nontrivial.…

Laparoscopic Field of View (FOV) control is one of the most fundamental andimportant components in Minimally Invasive Surgery (MIS) We present adata-driven framework to realize an automated laparoscopic optimal FOV control . We offline learn a motion strategy of laparoscoperelative to the surgeon’s hand-held surgical tool from our in-house surgicalvideos .…

Non-orthogonal multiple access (NOMA) is being widely considered as apotential candidate to enhance the spectrum utilization in beyond fifth-generation (B5G) communications . In contrast to the existing CRS-NomA systems, the system withreceive diversity outperforms its orthogonal (OMA-based) basedpart even in the low-SNR regime, by achieving higher ergodic rate .…

Capacity of a discrete-time model of optical fiber described by thesplit-step Fourier method (SSFM) as a function of the average input power$\mathcal P$ and the number of segments in distance $K$ is considered . The capacity of the resulting continuous-space lossless model is lower bounded by $o(1)$ .…

Code clones are duplicate code fragments that share (nearly) similar syntaxor semantics . Code clone detection plays an important role in software maintenance, code refactoring, and reuse . Researchers have attempted to adopt deep learning techniques to detect code clones .…

Magnetic coupling needs to be accounted forwhen designing memory arrays . Higher array densities show significant variations in average switching time, especially at low switching voltages, caused by inter-cell magnetic coupling . We also observe amarginal degradation of the data retention time under the influence ofinter-cellmagnetic coupling .…

We propose Kernel Predictive Control (KPC) is a learning-based predictivecontrol strategy that enjoys deterministic guarantees of safety . We present a relaxation strategy that exploits on-line datato weaken the optimization problem constraints while preserving safety . Twonumerical examples are provided to illustrate the applicability of the proposedcontrol method .…

This paper proposes a new model for augmenting algorithms with useful predictions that go beyond worst-case bounds on the algorithm performance . Byrefining existing models, our model ensures predictions are formally learnableand instance robust . Instance robustness formally ensures aprediction is robust to modest changes in the problem input .…

Earth system models (ESMs) are often used to generate future projections of climate change scenarios . Emulators are substantially less expensive but may not have all of the complexity of anESM . Here we demonstrate the use of a conditional generative adversarialnetwork (GAN) to act as an ESM emulator .…

We show howinterpersonalization can deliver engaging social visits in which visitors maketheir interpretations . We contrast the approach to previous research incustomization and algorithmic personalization . We propose that interpersonalization requires museums to step back to make space for interpretation, but that this then raises the challenge of how to reintroduce the museum’s own perspective .…

This letter considers a wireless network where an access point (AP)broadcasts timely updates to several mobile users . The timeliness ofinformation owned by a user is characterized by the recently proposed age ofinformation . While frequently broadcasting the timely updates and always using the maximum power can minimize the age of information for all users, that can waste valuable communication resources .…

LAX-RAY (Lateral Access maXimal Reduction of OccupancY support Area) is a system to automate the mechanical search for occluded objects on shelves . LAx-RAY couples a perception pipelinepredicting a target object occupancy support distribution with a mechanicalsearch policy that sequentially selects occluding objects to push to the sideto reveal the target as efficiently as possible .…

Planning is a powerful approach to reinforcement learning with several desirable properties . However, it requires a model of the world, which is not available in many real-life problems . In this paper, we propose tolearn a world model that enables Evolutionary Planning in Latent Space (EPLS) We use a Variational Auto Encoder (VAE) to learn a compressed latent latentrepresentation of individual observations and extend a Mixture DensityRecurrent Neural Network (MDRNN) The planning agents are better than standard model-freereinforcement learning approaches demonstrating the viability of our approach .…

We present a method for computing exact reachable sets for deep neuralnetworks with rectified linear unit (ReLU) activation . Our method is well-suited for use in rigorous safety analysis of robotic perception and control systems with deep neural network components .…

In this paper, we present a data-driven controller design method forcontinuous-time nonlinear systems with rational system dynamics . We use no model knowledge but only measured data affected by noise . We apply robust control techniques to this parametrization . We obtain sum-of-squares based criteria for designing controllers with closed-loop stability guarantees for all continuous-time systems withrational system dynamics which are consistent with the measured data and the assumed noise bound .…

In this paper, we present a new simple degree-based estimator for the size ofmaximum matching in bounded arboricity graphs . We show the estimator does better and returns a $3.5$ approximation of the matching size . Using this estimator, we get new results for approximating the matching sizes in the streaming and distributed models of computation .…

Real data often appear in the form of multiple incomplete views . Previous clustering methods only learn the consistent information between different views and ignore the unique information of each view . We propose a novel View Variation and View Heredity approach (V 3H) Inspired by the variation and the heredity in genetics, V 3H first decomposes each subspace into a variation matrix for the corresponding view and a redity matrix for all the views .…

A communication enabled indoor intelligent robots (IRs) service framework is proposed . Non-orthogonal multiple access (NOMA) technique is adopted to enable highly reliable communications . Lego modeling method is proposed, which can deterministicallydescribe indoor layout and channel state in order to construct the radiomap .…

A quadruped robot can acquire both of these skills by means of hierarchical reinforcementlearning (HRL) We train policies towalk in two different environments: a curved cliff and a maze . We show thathierarchical policies can concurrently learn to locomote and navigate in these environments .…

The proposed open-source Power Market Tool (POMATO) aims to enable research on interconnected modern and future electricity markets in the context of the transmission system and its secure operation . The computationally demanding N-k secure powerflow is enabled via an efficient constraint reduction algorithm .…

In an intelligent transportation system, the effects and relations of trafficflow at different points in a network are valuable features which can beexploited for control system design and traffic forecasting . In this work, directed information is used to determine the underlying graph structure of a network, denoted directedinformation graph, which expresses the causal relations among nodes in thenetwork .…

Radial basis function(RBF) interpolationmethods are known to approximate only functions in their native spaces . To date, there has been no known proof that they can approximate functions outsidethe native space associated with the particular RBF being used . In this paper, we describe a scattered data interpolation method which can approximate anyfunction of bounded variation from its scattered data as the data points growdense .…

A two-layer optimal control and operation scheme is proposed for a microgrid with energy storage . The working temperature of the battery is an important factor as a high-temperature condition generally increases losses, reduces useful life, and can even lead to fire hazards .…

Based on our Random Duality Theory (RDT), we developed a powerful algorithmic mechanism (called CLuP) that can be utilized to solve NP hard optimization problems in polynomial time . Here we move things further and utilize another ofremarkable RDT features that we established in a long line of work in the past .…

Graph neural networks (GNNs) are widely used in the applications based ongraph structured data, such as node classification and link prediction . Distance encoding (DE) (Li etal. 2020) magically makes GNNs work well in many applications, including nodeclassification and link predictions .…

We take commonly used QML ans\”{a}tze and convert them to diagrammatic form . We give a full description of how these gates commute . Weverage a combinatorial description of the interaction between CNOTs and phasegadgets to analyse a periodicity phenomenon in layered ans\’tze .…

Study addresses a longstanding concern in museum exhibition design that interactive technologies may distract from the artifacts on display . The Diary of Niels is an interactive installation that combines physical objects, RFIDsensors and an elaborate fiction in order to facilitate increased visitorengagement .…

Conventional methods often use semantics to colorizegrayscale images . In these methods, only classification semanticinformation is embedded, resulting in semantic confusion and color bleeding . To address these issues, we propose a fullyautomatic Saliency Map-guided Colorization with Generative Adversarial Network(SCGAN) framework .…

The application of Internet of Things (IoT) and Machine Learning (ML) to the agricultural industry has enabled the development and creation of smart farms . The growth in the number of smart farming has given rise to the CooperativeSmart Farming (CSF) where different connected farms collaborate with each other and share data for their mutual benefit .…

Recovery a planted signal perturbed by noise is a fundamental problem in machine learning . This fundamental problem appears in different contexts, e.g., community detection,average case complexity, and neuroscience applications . We provide tight information-theoretic upper and lowerbounds for the recovery problem, as well as the first non-trivial algorithmicbounds based on approximate message passing algorithms .…

Mixed regularized spectral clustering (Mixed-RSC) is an extension of the RSC method (Qin and Rohe, 2013) to deal with the mixed membership community detection problem . We show that the algorithm is asymptotically consistent under mild conditions . The approach is successfully applied to a small scale ofsimulations and substantial empirical networks with encouraging results compared to a number of benchmark methods .…

Fake news often involves semantic manipulations across modalities such asimage, text, location etc and requires the development of multimodal semanticforensics for its detection . The proposed model outperforms existing state-of-the-art algorithms with an error reduction of up to 25% . Existing methods arelimited to using a single evidence (retrieved package) which ignores potential improvement from the use of multiple evidences .…

This paper addresses the problem of restricted airspace protection frominvaders using the cooperative multi-UAV system . The objective is to detect and capture the invaders cooperatively by a team of homogeneous UAVs (calledevaders) before invaders enter the restricted airspace . The performance of MUST-MTA with the consensus approach is evaluated in a simulated environment .…