Due to recent advances in digital technologies, and availability of credibledata, an area of artificial intelligence, deep learning, has emerged, and hasdemonstrated its ability and... | Continue reading
Deep learning has been shown successful in a number of domains, ranging fromacoustics, images to natural language processing. However, applying deeplearning to the ubiquitous graph data is... | Continue reading
In this paper, we theoretically prove that we can eliminate all suboptimallocal minima by adding one neuron per output unit to any deep neural network,for multi-class classification, binary... | Continue reading
In this paper, we present a complete description of the hardware design andcontrol architecture of our custom built quadruped robot, called the Stoch. Ourgoal is to realize a robust, modular,... | Continue reading
Lots of learning tasks require dealing with graph data which contains richrelation information among elements. Modeling physics system, learningmolecular fingerprints, predicting protein... | Continue reading
Quantum key distribution (QKD) uses individual light quanta in quantumsuperposition states to guarantee unconditional communication security betweendistant parties. In practice, the achievable... | Continue reading
In a world where security issues have been gaining growing importance, facerecognition systems have attracted increasing attention in multiple applicationareas, ranging from forensics and... | Continue reading
We present a new hardware-agnostic side-channel attack that targets one ofthe most fundamental software caches in modern computer systems: the operatingsystem page cache. The page cache is a... | Continue reading
Deep learning has revolutionized many machine learning tasks in recent years,ranging from image classification and video processing to speech recognitionand natural language understanding. The... | Continue reading
The wavelet scattering transform is an invariant signal representationsuitable for many signal processing and machine learning applications. Wepresent the Kymatio software package, an... | Continue reading
Real-world face recognition requires an ability to perceive the uniquefeatures of an individual face across multiple, variable images. The primatevisual system solves the problem of image... | Continue reading
The growing literature on affect among software developers mostly reports onthe linkage between happiness, software quality, and developer productivity.Understanding the positive side of... | Continue reading
This paper describes the development of the Microsoft XiaoIce system, themost popular social chatbot in the world. XiaoIce is uniquely designed as an AIcompanion with an emotional connection to... | Continue reading
We present a traffic simulation named DeepTraffic where the planning systemsfor a subset of the vehicles are handled by a neural network as part of amodel-free, off-policy reinforcement learning... | Continue reading
Wireless devices are everywhere, at home, at the office, and on the street.Devices are bombarding us with transmissions across a wide range of RFfrequencies. Many of these invisible... | Continue reading
Learning in environments with large state and action spaces, and sparserewards, can hinder a Reinforcement Learning (RL) agent's learning throughtrial-and-error. For instance, following natural... | Continue reading
Remarks on the life and work of Paul Erdos. | Continue reading
We introduce Dynamic Planning Networks (DPN), a novel architecture for deepreinforcement learning, that combines model-based and model-free aspects foronline planning. Our architecture learns to... | Continue reading
AmbSAT is a biologically-inspired stochastic local search (SLS) solver toexplore solutions to the Boolean satisfiability problem (SAT). AmbSAT updatesmultiple variables in parallel at every... | Continue reading
CycleGAN (Zhu et al. 2017) is one recent successful approach to learn atransformation between two image distributions. In a series of experiments, wedemonstrate an intriguing property of the... | Continue reading
Keyword spotting with limited training data is a challenging task which canbe treated as a few-shot learning problem. In this paper, we present ameta-learning approach which learns a good... | Continue reading
In this work we present a theoretical model for differentiable programming.We present an algebraic language that enables both implementations and analysisof differentiable programs by way of... | Continue reading
A Wikipedia book (known as Wikibook) is a collection of Wikipedia articles ona particular theme that is organized as a book. We propose Wikibook-Bot, amachine-learning based technique for... | Continue reading
We propose a video compression framework using conditional GenerativeAdversarial Networks (GANs). We rely on two encoders: one that deploys astandard video codec and another which generates... | Continue reading
With cross-disciplinary academic interests increasing and academic advisingresources over capacity, the importance of exploring data-assisted methods tosupport student decision making has never... | Continue reading
We introduce an architecture to learn joint multilingual sentencerepresentations for 93 languages, belonging to more than 30 different languagefamilies and written in 28 different scripts. Our... | Continue reading
An implicit goal in works on deep generative models is that such modelsshould be able to generate novel examples that were not previously seen in thetraining data. In this paper, we investigate... | Continue reading
A detailed and rigorous analysis of Godel's proof of his first incompletenesstheorem is presented. The purpose of this analysis is two-fold. The first is toreveal what Godel actually proved to... | Continue reading
Social network and publishing platforms, such as Twitter, support the conceptof verification. Verified accounts are deemed worthy of platform-wide publicinterest and are separately authenticated... | Continue reading
The functionality that distributed ledger technology provides, i.e., animmutable and fraud-resistant registry with validation and verificationmechanisms, has traditionally been implemented with... | Continue reading
Practical challenges in simulating quantum systems on classical computershave been widely recognized in the quantum physics and quantum chemistrycommunities over the past century. Although many... | Continue reading
Let $F$ be a CM number field. We prove modularity lifting theorems forregular $n$-dimensional Galois representations over $F$ without anyself-duality condition. We deduce that all elliptic... | Continue reading
Machine learning has celebrated a lot of achievements on computer visiontasks such as object detection, but the traditionally used models work withrelatively low resolution images. The... | Continue reading
Gesture and signature passwords are two-dimensional figures created bydrawing on the surface of a touchscreen with one or more fingers. Prior resultsabout their security have used resilience to... | Continue reading
We introduce the use of high order automatic differentiation, implemented viathe algebra of truncated Taylor polynomials, in genetic programming. Using theCartesian Genetic Programming encoding... | Continue reading
A quantum computer (QC) can solve many computational problems moreefficiently than a classic one. The field of QCs is growing: companies (such asDWave, IBM, Google, and Microsoft) are building... | Continue reading
We report a new model of artificial life called Lenia (from Latin lenis"smooth"), a two-dimensional cellular automaton with continuousspace-time-state and generalized local rule. Computer... | Continue reading
In this paper, we propose a new deep neural network architecture, called NMDnet, that has been specifically designed to learn adaptive behaviours. Thisarchitecture exploits a biological... | Continue reading
We present Wikipedia2Vec, an open source tool for learning embeddings ofwords and entities from Wikipedia. This tool enables users to easily obtainhigh-quality embeddings of words and entities... | Continue reading
A recent paper [31] claims to classify brain processing evoked in subjectswatching ImageNet stimuli as measured with EEG and to use a representationderived from this processing to create a novel... | Continue reading
Sized types are a modular and theoretically well-understood tool for checkingtermination of recursive and productivity of corecursive definitions. Theessential idea is to track structural... | Continue reading
The aggregation of many independent estimates can outperform the mostaccurate individual judgment. This centenarian finding, popularly known as thewisdom of crowds, has been applied to problems... | Continue reading
Software development includes diverse tasks such as implementing newfeatures, analyzing requirements, and fixing bugs. Being an expert in thosetasks requires a certain set of skills, knowledge,... | Continue reading
The use of Project Gutenberg (PG) as a text corpus has been extremely popularin statistical analysis of language for more than 25 years. However, incontrast to other major linguistic datasets of... | Continue reading
3D object detection is an essential task in autonomous driving. Recenttechniques excel with highly accurate detection rates, provided the 3D inputdata is obtained from precise but expensive... | Continue reading