Transfer Learning for Human & AI

We propose an online heuristic algorithm for the problem of network sliceplacement optimization . The solution is adapted to support placement on largescale networks and integrates Edge-specific and URLLC constraints . We rely on an approach called the Power of Two Choices to build the heuristic .…

Domain-specific languages (DSLs) are both pervasive and powerful, but remain difficult to integrate into large projects . We describe an extension to the Clang C++ compiler to support syntax plugins . We demonstrate how this mechanism allows making use of DSLs insideof a C++ code base without needing to separate the DSL source code from the surrounding C++code .…

The feedback motion planning algorithm uses a quasi-Newton method to simultaneously optimize a nominaltrajectory, a set of deterministically chosen sample trajectories, and aparameterized policy . We demonstrate that this approach exactly recovers LQRpolicies in the case of linear dynamics, quadratic cost, and Gaussian noise .…

Global Navigation Satellite System (GNSS) provides Positioning, Navigation,and Timing (PNT) services for autonomous vehicles (AVs) A spoofed attack can mimic the GNSS signal and transmit inaccurate location coordinatesto an AV . In this study, we have developed a prediction-based spoofing attack detection strategy using the long short-term memory (LSTM) model, a recurrentneural network model .…

Imperative session types provide an imperative interface to session-typed communication in a functional language . Compared to functional session typeAPIs, the program structure is simpler at the surface, but typestate is required to model the current state of communication throughout .…

In this demo paper, we consider the network slice placement optimizationproblem and give some insights into a fast heuristic algorithm tailored toplacement in large scale networks . We consider an online optimization scenariowith multiple and volatile network slice request arrivals and we showcase theapplicability of the proposed Edge-enabled network slice placements solution .…

Imitation learning from observation is more preferable than imitationlearning from demonstration (LfD) due to the nonnecessity of expert actions . Previous studies imply that the performance of LfO is inferior to LfD by a tremendous gap, which makes it challenging to employ LfF in practice .…

Manoeuvre Intelligence for SpaceSafety (MISS) is a new software tool for collision avoidance analysis and design . The tool leverages analytical and semi-analytical methods for the efficient modelling of the orbit modifications due to different controlstrategies, such as impulsive or low-thrust manoeuvres, and maps them intodisplacements at the nominal close approach using relative motion equations .…

The task of hemostasis covers aspectrum of bleeding sources and a range of blood velocity, trajectory, and volume . In an extreme case, an un-controlled blood vessel fills the surgicalfield with flowing blood . The proposed methods are powerful tools to clearing the surgical field which can be followed by either asurgeon or future robotic automation developments to close the vessel rupture .…

This paper presents a formal verification guided approach for a . principleddesign and implementation of robust and resilient learning-enabled systems . The robustness verification is NP-complete, according to the authors . We investigate two key properties – robustness and resilience – and provide their formal definitions .…

Alocal BA on a sliding window of keyframes has been widely used in visual SLAMand has proved to be very effective in lowering the drift . But in lidarmapping, BA method is hardly used because the sparse feature points (e.g., edgeand plane) in a lidar point-cloud make the exact feature matching impossible .…

Accurate post-impact velocity predictions are essential in developing impact-aware manipulation strategies for robots, where contacts are intentionally established at non-zero speed mimicking human manipulationabilities in dynamic grasping and pushing of objects . The results of this new approach arepromising in terms of prediction accuracy and thus relevant for the growingfield of impact-awareness robot control .…

Accurately modeling local surface properties of objects is crucial to manyrobotic applications, from grasping to material recognition . Surface properties like friction are difficult to estimate, as visual observation of the object does not convey enough information over these properties .…

Few-shot adaptation is a challenging problem in the context of simulation-to-real transfer in robotics, requiring safe and informative datacollection . In physical systems, additional challenge may be posed by domainnoise, which is present in virtually all real-world applications . The proposed method also allows for system analysis by analyzing hidden states of the model during and after adaptation .…

Probabilistic Programming Languages (PPLs) are a powerful tool in machinelearning, allowing highly expressive generative models to be expressed . They couple complex inference algorithms, implemented by the language, with an expressive modelling language that allows a user to implementany computable function as the generative model .…

Real-world systems often involve external interventions that cause changes in the system dynamics such as a moving ball coming in contact with another ball . We propose a novel neural ODE-based approach (IMODE) that properly model the effect of external interventions by employing two ODE functions to handle the observations and the interventions separately .…

This paper studies a new Whitney type inequality on a compact domain$\Omega\subset {\mathbb{R}}^d$ . It is proved that $d=2$ and $d =2$ for every connected$C^2$- domain $D$ and every planar convexbody $d$ is proved .…

Photoacoustic tomography aims at measuring the high-contrast optical properties of tissues by means of ultrasonic measurements . The interaction between twotypes of waves is based on the thermoacoustic effect . In many measurement setups of practical interest, compressedsensing PAT reduces to compressed sensing for undersampled Fouriermeasurements.…

Data assimilation algorithms combine prior and observational information to obtain the most likely posterior of a dynamical system . Many numerical weatherprediction centres use full observation error covariance (OEC) weightingmatrices, which can slow convergence of the data assimilation procedure . We find that the minimumeigenvalue of the OECmatrix appears in bounds on the condition number of the Hessian of the preconditioned objective function .…

This paper presents a computationally efficient and robust LiDAR-inertialodometry framework . The proposed method and its implementation are tested in various indoor and outdoor environments . In all tests, our method produces reliablenavigation results in real-time: running on a quadrotor onboard computer, it .…

In this paper, we present a novel two-level variable Horizon Model PredictiveControl (VH-MPC) framework for bipedal locomotion . In this framework, thehigher level computes the landing location and timing (horizon length) of the swing foot to stabilize the unstable part of the center of mass (CoM) dynamics .…

In order to enable robust operation in unstructured environments, robotsshould be able to generalize manipulation actions to novel object instances . With ACF, we represent each object class interms of individual affordance parts and the compatibility between them . We demonstrate that ACF outperforms state-of-the-artmethods for object detection, as well as category-level pose estimation for robot manipulation .…

In typical MBRL, we cannot expect the data-driven model to generate accurate and reliable policies to the intended robotic tasks during the learning process due to data scarcity . Operating these unreliable policies in acontact-rich environment could cause damage to the robot and its surroundings .…

Successful robotic operation in stochastic environments relies on accuratecharacterization of the underlying probability distributions . This work presents a control algorithm that is capable of handling such distributional mismatches . The benefits of thedistributional robustness as well as the automatic risk-sensitivity adjustmentare demonstrated in a dynamic collision avoidance scenario where the predictivedistribution of human motion is erroneous .…

Multi-agent path finding (MAPF) is an indispensable component of large-scalerobot deployments in numerous domains ranging from airport management to warehouse automation . We present extensive resultsof PRIMAL2 in both MAPF and LMAPF environments with up to 1024 agents and compare its performance to complete state-of-the-art planners .…

Using a moment optimization approach, we formulate ahierarchy of semidefinite programs that yield increasingly refined lower bounds . We propose a novel method for planning shortest length piecewise-linearmotions through complex environments . Our method natively handles continuoustime constraints without any need for time discretization, and has the potential to scale better with dimensions compared to popular sampling-based methods .…

Learning from Hallucination (LfH) is a recent machine learning paradigm for autonomous navigation . It uses training data collected in completely safe environments and adds imaginary obstacles to make the environment densely constrained . LfH requires hallucinating the robot perception during deployment to match with the training data, which creates a need for sometimes-infeasible priorknowledge and tends to generate very conservative planning .…

Panoptic Segmentation aims to provide an understanding of background (stuff)and instances of objects (things) at a pixel level . It combines the separatetasks of semantic segmentation (pixel-level classification) and instancesegmentation to build a single unified scene understanding task . In this work, we present a fully convolution neural network that learns instance segmentation from semantic and instance contours (boundaries of things) and labels .…

Planning under uncertainty is a crucial capability for autonomous systems tooperate reliably in uncertain and dynamic environments . The concern of patientsafety becomes even more critical in healthcare settings where robots interact with humans . In this paper, we propose a novel risk-aware planning framework tominimize the risk of patient falls by providing a patient with an assistivedevice .…

Adaptive featureselection (AFS) for encoder-decoder based E2E ST . We pre-train an ASRencoder and apply AFS to dynamically estimate the importance of each encodedspeech feature to SR . We take L0DROP(Zhang et al., 2020) as the backbone for AFS, and adapt it to sparsify speechfeatures with respect to both temporal and feature dimensions .…

This paper analyses the robust stability and performance of the DisturbanceObserver- (DOb-) based digital motion control systems in discrete-time domain . It is shown that the phase margin and the robustness of the digital motioncontroller can be directly adjusted by tuning the nominal plant model and thebandwidth of the observer .…

Sound event detection (SED) and acoustic scene classification (ASC) are important research topics in environmental sound analysis . Many research groups have addressed SED and ASC using neural-network-based methods . In this paper, we propose multitask learning for joint analysis of sound events and acoustic scenes, in which the parts of thenetworks holding information on sound events in common are shared .…

Given an unknown audio source, the estimation of time differences-of-arrivals(TDOAs) can be efficiently and robustly solved using blind channel identification and exploiting the cross-correlation identity (CCI) We seek to warm up the community and the practitioners by paving theway (with two concrete, yet preliminary, examples) towards joint approaches in which advances in the optimization are combined with an increased number of microphones, in order to achieve further improvements .…

Programming languages development has intensified in recent years . New features, often cross-paradigm, are featured in old ones . This new programming landscape makes language selection a more complex decision . In this paper, however, we arguethat programming languages have a secondary role in software development .…

Impersonators are playing an important role in the production and propagation of the content on Online Social Networks, notably on Instagram . These entities are nefarious fake accounts that intend to disguise a legitimate account by making similar profiles and then striking social media by posting fake content .…

This paper examines the fundamental problem of identifying the most important nodes in a network . We propose six simple properties and prove that PageRank is the only centrality measure that satisfies all of them . Our work gives new conceptual andoretical foundations of PageRank that can be used to determine suitability of this measure in specific applications .…

Under preferential attachment (PA) network growth models late arrivals are at a disadvantage with regard to their final degrees . Here we introduce a new dynamical approach to address late arrivals by adding a recent-degree-change bias to PA so that nodes with higher relative degree change in temporal proximity to an arriving node get an attachment probability boost .…

Huntington disease (HD) is a fatal autosomal dominant neurocognitive disorder . It causes cognitive disturbances, neuropsychiatric symptoms, and impaired motor abilities (e.g., gait, speech, voice) Individuals with HD may exhibit a range of speech symptoms . Speech-based passivemonitoring has the potential to augment clinical assessments by continuouslytracking manifestation symptoms .…

Inferring graph structure from observations on nodes is an important and popular network science task . We study theproblem of jointly inferring multiple graphs from the observation of signals attheir nodes . From a mathematical point of view, graph stationarity implies that themapping between the covariance of the signals and the sparse matrixrepresenting the underlying graph is given by a matrix polynomial .…

Political astroturfing and organised trolling are online malicious behaviour with significant real-world effects . Common approaches examining these phenomena focus on broad campaigns rather than small groups responsible . We propose a novel temporalwindow approach that relies on account interactions and metadata alone .…

The presidential elections in the United States on 3 November 2020 havecaused extensive discussions on social media . A part of the content on USelections is organic, coming from users discussing their opinions of the candidates, political positions, or relevant content presented on television .…

Solving optimization tasks based on functions and losses with a topological flavor is a growing field of research in Topological DataAnalysis . Most of these methods rely on the fact that most of the topological constructions are stratifiable and differentiable almost everywhere .…

We consider a railway dispatching problem: delay and conflict management on asingle-track railway line . We examine the issue of train dispatchingconsequences caused by the arrival of an already delayed train to the segment being considered . This is a computationally hard problem and its solution is needed in a very short time in practice .…

The AD model is a mimic of the attention diffusion process along the citation networks . For some newly published papers that have not accumulated many citations butwill be popular in the near future, the AD model can substantially improve their rankings .…

Philosophers of science have long postulated how collective scientificknowledge grows . Empirical validation has been challenging due to limitations in collecting and systematizing large historical records . We capitalizeon the largest online encyclopedia to formulate knowledge as growing networksof articles and their hyperlinked inter-relations .…

High-speed rail (HSR) has become an important mode of inter-city transportation between large cities . Quantifying the impact of HSR’s interaction on cities and people is crucial for long-term urban and regional planning and policy making . The results show that HSRs notonly greatly expand the uneven distribution of inter city connections, but alsosignificantly reshape the interactions that occur along HSR routes through the channel effect .…

The Automatic Speaker Verification Spoofing and Countermeasures Challengesmotivate research in protecting speech biometric systems against a variety of different access attacks . The 2017 edition focused on replay spoofing attacks, and involved participants building and training systems on a provided dataset(ASVspoof 2017) More than 60 research papers have so far been published with the dataset .…

This paper proposes a framework for 3D obstacle avoidance in the presence ofpartial observability of environment obstacles . The method focuses on theutility of the Artificial Potential Field (APF) controller in a practical setting where noisy and incomplete information about the proximity isinevitable .…

Mandarin speech recordings from a Mandarin speaker and English from another English speakerto build high-quality bilingual and code-switched TTS for both speakers . ATacotron2-based cross-lingual voice conversion system is employed to generatethe Mandarin speaker’s English speech and the English speaker’s Mandarinspeech, which show good naturalness and speaker similarity .…

Multi-agent reinforcement learning has recently shown great promise as an approach to networked system control . Here we take inspiration from a recent research program investigating the game-theoretic incentives of humans in social dilemma situations such as the well-known tragedy of thecommons .…