Sequential Resource Access Theory and Algorithm

We formulate and analyze a generic sequential resource access problem arising in a variety of engineering fields . We present a greedy strategy implementable in linear time . We develop a series of polynomial-time approximation algorithms achieving $(\epsilon,\delta)-optimality . The key component of the algorithm is a pruning processeliminating dominated strategies and, thus maintaining polynnomine time and space overhead.…

The local discontinuous Galerkin method on layer adapted meshes for time dependent singularly perturbed convection diffusion problems

In this paper we analyze the error as well for the semi-discretization of a time-dependent convection-diffusion problem . We usefor the discretization in space the local discontinuous Galerkin (LDG) method . Our analysis is based on careful approximationerror estimates for the Ritz projection related to the stationary problem on the anisotropic meshes used .…

Nonnegative Matrix Factorization with Toeplitz Penalty

Nonnegative Matrix Factorization (NMF) is an unsupervised learning algorithm that produces a linear, parts-based approximation of a data matrix . NMFconstructs a nonnegative low rank basis matrix and a nonzero low rank matrix of weights which, when multiplied together, approximate the data matrix of interest using some cost function .…

Efficient Reservoir Management through Deep Reinforcement Learning

Dams impact downstream river dynamics through flow regulation and disruption of upstream-downstream linkages . The unsatisfactory dam operation can cause floods in downstream areas . We leverage reinforcement learning (RL) methods to compute dam operation guidelines . We build offlinesimulators with real data and different mathematical models for the upstreaminflow, i.e.,…

HEBO Heteroscedastic Evolutionary Bayesian Optimisation

HEBO: Heteroscedastic Evolutionary Evolutionary Bayesian Optimisation that won NeurIPS 2020 black-box optimisation competition . We presentnon-conventional modifications to the surrogate model and acquisitionmaximisation process . We perform an ablation study to highlight the components that contributed to the success of HEBO .…

Nonnegative Matrix Factorization with Zellner Penalty

Nonnegative matrix factorization (NMF) decomposes a nonnegative data matrix into an aparts-based, lower dimensional, linear representation of the data . NMF has applications in image processing, text mining, recommendation systems and avariety of other fields . We assess the facialrecognition performance of the ZNMF algorithm and several other well-known NMF algorithms using the Cambridge ORL database .…

Multitask machine learning of collective variables for enhanced sampling of rare events

A data-driven machine learning algorithm is devised to learn collective variables with a multitask neuralnetwork . A common upstream part reduces the high dimensionality of atomicconfigurations to a low dimensional latent space . The resulting latent space is shown to be an effectivelow-dimensional representation, capturing the reaction progress and guiding umbrella sampling to obtain accurate free energy landscapes .…

Exploiting Implicit Kinematic Kernel for Controlling a Wearable Robotic Extra finger

The novel idea here presented lies on taking advantage of the redundancy of the human kinematic chain involved in a task to control an extra degree of actuation (DoA) This concept is summarized in thedefinition of the Implicit Kinematic Kernel (IKK) We considered both bio-mechanical and physiological human features and constraints to allow for an efficient and intuitive control approach .…

On Infusing Reachability Based Safety Assurance within Planning Frameworks for Human Robot Vehicle Interactions

Key challenge is accounting for uncertainty in human driver actions without unduly impacting planner performance . Paper introduces aminimally-interventional safety controller operating within an autonomous vehicle control stack with the role of ensuring collision-free interaction with an externally controlled (e.g., human-driven) counterpart while respecting static obstacles such as a road boundary wall .…

Small Changes Big Impacts Leveraging Diversity to Improve Energy Efficiency

A growing body of research has proposed methods,techniques, and tools to support developers in the construction of software that consumes less energy . We advocate thatdevelopers should leverage software diversity to make software systems moreenergy-efficient . Non-specialists can build software by alternating at development time between readily available, diversely-designed pieces of software implemented by third-parties .…

Source Separation and Depthwise Separable Convolutions for Computer Audition

We train a depthwise separable convolutionalneural network on a challenging electronic dance music data set . It is shown that source separation improves classification performance in a limited-data setting compared to the standard single spectrogram approach . We propose a featurerepresentation method that combines source separation with state-of-the-artrepresentation learning technique that is suitably repurposed for computeraudition (i.e.…

EfficientTTS An Efficient and High Quality Text to Speech Architecture

In this work, we address the Text-to-Speech task by proposing anon-autoregressive architecture called EfficientTTS . The proposed models significantly outperform counterpart models such as Tacotron 2 and Glow-TTS in terms of speech quality, training efficiency and training efficiency . In addition, we demonstrate that proposed approach can beeasily extended to autoregressive models .…

Data driven Model Predictive Control Method for DFIG based Wind Farm to Provide Primary Frequency Regulation Service

As wind power penetration increases, wind farms are required by grid codes to provide frequency regulation service . The most critical challenge is how to formulate the dynamic model of wind farm for dynamiccontrol . This paper proposes a data-driven modelpredictive control (data-driven MPC) method to make wind farms participate primary frequency regulation .…

NaturalCC A Toolkit to Naturalize the Source Code Corpus

We present NaturalCC, an efficient and extensible toolkit to bridge the gap between natural language and programming language . Using NaturalCC researchers can quickly and easily reproduce the state-of-the-art baselines and implement their approach . The video of this demo is available athttps://www.y.com/watch?v=q4W5VSI-u3E&t=25s.…

An Improved Benders Decomposition Algorithm for Steady State Dispatch Problem in an Integrated Electricity Gas System

Optimally operating an integrated electricity-gas system (IEGS) issignificant for the energy sector . The IEGS operation model’snonconvexity makes it challenging to solve the optimal dispatch problem . This letter proposes an improved Benders decomposition (IBD) algorithm . Case studies substantiated the higher efficiency of our IBD algorithm, which leverages a refined decomposition structure where the subproblemsbecome linear and ready to be solved in parallel .…

Space Filling Subset Selection for an Electric Battery Model

Dynamic models of the battery performance are an essential tool throughout the development process of automotive drive trains . The present studyintroduces a method making a large data set suitable for modeling theelectrical impedance . The algorithm selectsthose dynamic data points that fill the input space of the nonlinear model morehomogeneously .…

Near Real Time Social Distancing in London

Policy makers at the Greater London Authority are reliant upon prompt and accurate data sources during the COVID-19 pandemic . Our method enables near immediate sampling and contextualisation of activity and physical distancing on the streets of London via live trafficcamera feeds .…

Data Driven Predictive Control for Continuous Time Industrial Processes with Completely Unknown Dynamics

The proposed approach employs the data-driven technique to get the systemmatrices online, using input-output measurements . Then, a model-free predictivecontrol approach is designed to implement the receding-horizon optimization andrealize the reference tracking . Feasibility of the proposed algorithm and stability of the closed-loop control systems are analyzed .…

Improved Convergence Rates for Non Convex Federated Learning with Compression

Federated learning is a new distributed learning paradigm that enablesefficient training of emerging large-scale machine learning models . In thispaper, we consider federated learning on non-convex objectives with compressedcommunication from the clients to the central server . The proposed scheme is the first algorithm that attains theaforementioned optimal complexity with compressed communication and withoutusing full client gradients at each communication round .…

Codimensional Incremental Potential Contact

We extend the incremental potential contact (IPC) model [Li et al. 2020a] to resolve systems composed of arbitrary combinationsof codimensional degrees-of-freedoms . This enables a unified,interpenetration-free, robust, and stable simulation framework that couplescodimension-0,1,2, and 3 geometries seamlessly with frictional contact .…

Galloping in natural merge sorts

We study the algorithm TimSort and the sub-routine it uses to merge monotonic(non-decreasing) sub-arrays . We look at the impact on the number of element comparisons performed of using thissub-ranoutine instead of a naive routine . In this article, we introduce a new object for measuring the complexity ofarrays.…

Quality Estimation Interpretability for Code Translation

The code translation task is an analog of machine translation (MT) for natural languages, with some added caveats . In this paper, we attempt to estimatethe quality of source code translations built on top of the TransCoder model . We present our main motivationfrom a user study built around code translation; and present a technique thatrelates the confidences generated by that model to lint errors in the translated code .…

Modeling and computer simulation of the mixing and heat transfer in heterogeneous turbulent two phase jets of mutually immiscible liquids by the method of Professor Alfred I Nakorchevskii Part 1

Many natural and technical processes deal with the turbulent jets of mutuallyimmiscible liquids . They represent an important class of the modern multiphasesystem dynamics . Differential equations for the axially symmetrical two-dimensional stationary flow and the integral correlations in a cylindrical coordinate system were considered for the jet from a nozzle into a space filled with another fluid that is not miscible with the first one .…

TediGAN Text Guided Diverse Image Generation and Manipulation

In this work, we propose TediGAN, a novel framework for multi-modal imagegeneration and manipulation with textual descriptions . The proposed method consists of three components: StyleGAN inversion module, visual-linguistics similarity learning, and instance-level optimization . Our model can provide the lowest effect guarantee, and producediverse and high-quality images with an unprecedented resolution at 1024 .…

Cost effective Machine Learning Inference Offload for Edge Computing

Computing at the edge is increasingly important since a massive amount of data is generated . This poses challenges in transporting all that data to theremote data centers and cloud, where they can be processed and analyzed . But harnessing the edge data is essential for offering data-driven andmachine learning-based applications, if the challenges, such as devicecapabilities, connectivity, and heterogeneity can be mitigated .…

Stellar Resolution Multiplicatives for the linear logician through examples

The stellar resolution is an asynchronous model of computation used in Girard’s Transcendental Syntax . It is based on Robinson’s first-order clausalresolution . By using methods of realisability for linear logic, we obtain a new model of multiplicative linear logic (MLL) Based on sort of logic programscalled constellations which are used to represent proofs, cut-elimination, and correctness and provability very naturally .…