UAV Localization Using Autoencoded Satellite Images

Previous work using satellite images has large storage and computation costs and is unable to run in realtime . In this work, we collect Google Earth (GE) images for a desired flightpath offline and an autoencoder is trained to compress these images to alow-dimensional vector representation while retaining the key features .…

Systematic Generalization for Predictive Control in Multivariate Time Series

In our context, systematic generalization implies that a good modelshould perform well on all new combinations of future actions after being trained on all of them, but only on a limited set of combinations . We conduct a rigorous study of useful inductive biases that learn topredict the trajectories up to large horizons well, and capture true dependencyrelations between the states and the controls through our synthetic setup, and simulated data from electric motors .…

Belief Space Planning for Mobile Robots with Range Sensors using iLQG

In this work, we use iterative Linear Quadratic Gaussian to plan plans for a mobile robot with range sensors in belief space . We address twolimitations that prevent applications of iLQG to the considered robotic system . We propose to use a derivative-free filter to approximate thebelief dynamics, which does not require explicit differentiability of themeasurement model .…

Manipulability optimization for multi arm teleoperation

Teleoperation provides a way for human operators to guide robots in situations where full autonomy is challenging or where direct human intervention is required . The increased availability ofcollaborative robot arms and Virtual Reality (VR) devices provides ampleopportunity for development of novel teleoperation methods .…

Learning Interaction Aware Trajectory Predictions for Decentralized Multi Robot Motion Planning in Dynamic Environments

This paper presents a data-driven trajectory optimization approach for multi-robot motion planning in dynamic environments . Whennavigating in a shared space, each robot needs accurate motion predictions ofneighboring robots to achieve predictive collision avoidance . These motionpredictions can be obtained among robots by sharing their future plannedtrajectories with each other via communication .…

Origami spring inspired shape morphing for flexible robotics

Flexible robotics are capable of achieving various functionalities by shapemorphing, benefiting from their compliant bodies and reconfigurable structures . The proposed concept of origami-aideddesign is expected to pave the way to facilitate diverse shape morphing offlexible robotics . To improve the mechanical performance such asthe damage resistance, we establish an origami rigidization method by adding additional creases to the spring system .…

Polarimetric Monocular Dense Mapping Using Relative Deep Depth Prior

This paper is concerned with polarimetric dense map reconstruction based on apolarization camera with the help of relative depth information as a prior . In general, polarization imaging is able to reveal information about surfacenormal such as azimuth and zenith angles, which can support the development of solutions to the problem of dense reconstruction, especially in texture-poor regions .…

Dompteur Taming Audio Adversarial Examples

In our digital society, this poses a significant threat to machine learning systems . We accept the presence of adversarial examples against ASR systems, but we require them to be perceivable by human listeners . By applying the principles of psychoacoustics, we can remove semantically irrelevant information from the ASR input and train a model that resembles human perception more closely .…

Sound Event Detection Based on Curriculum Learning Considering Learning Difficulty of Events

In conventional sound event detection (SED) models, two types of events are regarded as the same type of events . We propose a new objective function for SED, wherein the events are trained from easy- to difficult-to-train events . Experimental results show that the F-score of the proposed method is improved by 10.09 percentage points compared with that of the conventional binary cross entropy-based SED .…

GuiltyWalker Distance to illicit nodes in the Bitcoin network

Money laundering is a global phenomenon with wide-reaching social and economic consequences . Cryptocurrencies are particularly susceptible due to the lack of control by authorities and their anonymity . We propose new features based on the structure of the graph and past labels to boost the performance of machine learning methodsto detect money laundering .…

Safe visor Architecture for Sandboxing AI based Unverified Controllers in Stochastic Cyber Physical Systems

High performance but unverified controllers, e.g., artificialintelligence-based (a.k.a. AI-based) controllers, are widely employed incyber-physical systems (CPSs) to accomplish complex control missions . The proposed architecture contains a history-based supervisor, which checks inputs from theunverified controller, and a safety advisor that provides fallback when the unverified controller endangers the safety of the system .…

Adaptive Processor Frequency Adjustment for Mobile Edge Computing with Intermittent Energy Supply

Mobile Edge Computing (MEC) has played an increasingly important role in the next generation of connectivityand service delivery . Yet, along with the massive deployment of MEC servers, the ensuing energy issue is now on an increasingly urgent agenda . In the current context, the large scale deployment of renewable-energy-supplied MECservers is perhaps the most promising solution for the incoming energy issue .…

Multi Scale Neural Networks for to Fluid Flow in 3D Porous Media

The permeability of complex porous materials can be obtained via direct flowsimulation, which provides the most accurate results, but is very expensive . Data-driven machine learning approaches have shown great promise for building more general models . However, prior approaches building on Convolutional NeuralNetwork (ConvNet) literature concerning 2D image recognition problems do not scale well to the large 3D domains required to obtain a RepresentativeElementary Volume .…

On the Power and Limitations of Branch and Cut

The Stabbing Planes proof system is powerful enough to simulate Cutting Planes and refute the Tseitinformulas . Dadush and Tiwari showed that these short refutations of these formulas could be translated into quasi-polynomial size and depth Cutting Planes proofs, refuting a long-standing conjecture .…

Constant Approximating k Clique is W 1 hard

On input a graph $G$ a positive integer $k$ and a small constant $epsilon0$ is a simple algorithm . The algorithm can distinguish between the cases$\omega(G)$ and $f(k/c$ for any constant $c\ge 1$ and computablefunction $f$ unless $FPT= W[1]$. This implies that no$f(K))\cdot |G|^{O(1)$-time algorithm can .…

Structure vs Randomness for Bilinear Maps

We prove that the slice rank of a 3-tensor is equivalent up to anabsolute constant . We obtain strong trade-offs on the complexity of a biased bililnear map . Our result settles open questions of Haramaty and Shpilka [STOC 2010], and of Lovett [Discrete Anal.,2019]…

Structure preserving Model Reduction of Parametric Power Networks

We develop a structure-preserving parametric model reduction approach for swing equations . We employ a global basis approach to develop theparametric reduced model . We concatenate the local bases obtained via $\mathcal{H}_2-based interpolatory model reduction . The residue of theunderlying dynamics corresponding to the simple pole at zero varies with theparameters .…

Smart non intrusive appliance identification using a novel local power histogramming descriptor with an improved k nearest neighbors classifier

Non-intrusive load monitoring (NILM) is a key cost-effective technology formitoring power consumption and contributing to several challenges encountered . This paper proposes a smart NILM system based on a novel localpower histogramming (LPH) descriptor, in which appliance power signals aretransformed into 2D space and short histograms are extracted to represent each device .…

Securing the Network for a Smart Bracelet System

Digital instruments play a vital role in our daily life . Security is essential for healthcare systems as the blood pressure recordings by the smart bracelet are sent to the user’s mobile phone via Bluetooth . The bracelet monitors the pregnant women, but also other users whowish to have their blood pressure under control .…

Interrogating the Black Box Transparency through Information Seeking Dialogues

This paper is preoccupied with the question: given a (possiblyopaque) learning system, how can we understand whether its behaviour adheres togovernance constraints? The answer can be quite simple: we just need to “ask” the system about it . We propose to construct an investigator agent to query a learning agent to investigate its adherence to a given ethical policy in the context of an information-seeking dialogue, modeled informal argumentation settings .…

DHLink A Microservice Platform supporting Rapid Application Development and Secure Real time Data Sharing in Digital Health

Digital health applications that leverage multiple sources of patient data are becoming increasingly popular invarious medical practices and research . DHLink securely links existing digitalhealth applications of different projects, facilitates real-time data sharing, and supports rapid application development by reusing data and functions of existing digital health applications .…

A High Performance Sparse Tensor Algebra Compiler in Multi Level IR

Tensor algebra is widely used in many applications, such as scientific computing, machine learning, and data analytics . The performance of sparsetensor operations depends on a particular architecture and/or selected sparseformat . We propose a tensor algebra domain-specific language (DSL) and compiler infrastructure to automatically generate kernels for mixed sparse-dense tensoralgebra operations, named COMET .…

OptSmart A Space Efficient Optimistic Concurrent Execution of Smart Contracts

Popular blockchains such as . Ethereum execute complextransactions in blocks through user-defined scripts known as smart contracts . We rigorously prove the correctness of concurrentexecution of AUs and achieve significant performance gain over the state-of-the-art . We propose aconcurrent validator that re-executes the same AUs concurrently anddeterministically using a concurrent bin followed by a BG given by the miner to .verify…