TensorNetwork is an open source library for implementing tensor networkalgorithms. Tensor networks are sparse data structures originally designed forsimulating quantum many-body physics, but are... | Continue reading
Molecular data systems have the potential to store information atdramatically higher density than existing electronic media. Some of the firstexperimental demonstrations of this idea have used... | Continue reading
Recent work on the "lottery ticket hypothesis" proposes thatrandomly-initialized, dense neural networks contain much smaller, fortuitouslyinitialized subnetworks ("winning tickets") capable of... | Continue reading
Somewhat surprisingly, the behavior of analytical query engines is cruciallyaffected by the dynamic memory allocator used. Memory allocators highlyinfluence performance, scalability, memory... | Continue reading
This paper aims to improve upon the generic recommendations that Redditprovides for its users. We propose a novel personalized recommender system thatlearns from both, the presence and the... | Continue reading
Artificial Intelligence (AI), defined in its most simple form, is atechnological tool that makes machines intelligent. Since learning is at thecore of intelligence, machine learning poses itself... | Continue reading
We present Elmo, a system that addresses the multicast scalability problem inmulti-tenant data centers. Modern cloud applications frequently exhibitone-to-many communication patterns and, at the... | Continue reading
The common perception in both academic literature and the industry today isthat virtual machines offer better security, while containers offer betterperformance. However, a detailed review of... | Continue reading
The Keynesian Beauty Contest is a classical game in which strategic agentsseek to both accurately guess the true state of the world as well as theaverage action of all agents. We study an... | Continue reading
Background. Expert observers and artificial intelligence (AI) progressionmetrics indicate AI will exceed human intelligence within a few decades.Whether general AI that exceeds human... | Continue reading
We present a method for automatically organizing and evaluating the qualityof different publishing venues in Computer Science. Since this method onlyrequires paper publication data as its input,... | Continue reading
With this paper, we survey the research performed in the domain of browserfingerprinting, while providing an accessible entry point to newcomers in thefield. We explain how this technique works... | Continue reading
Domain generation algorithms (DGAs) are commonly leveraged by malware tocreate lists of domain names which can be used for command and control (C&C)purposes. Approaches based on machine learning... | Continue reading
Humans gather information by engaging in conversations involving a series ofinterconnected questions and answers. For machines to assist in informationgathering, it is therefore essential to... | Continue reading
Voluntary insurance contracts constitute a puzzle because they increase theexpectation value of one party's wealth, whereas both parties must sign forsuch contracts to exist. Classically, the... | Continue reading
This paper addresses the challenging problem of retrieval and matching ofgraph structured objects, and makes two key contributions. First, wedemonstrate how Graph Neural Networks (GNN), which... | Continue reading
A new set of DNA base-nucleic acid codes and their hypercomplex numberrepresentation have been introduced for taking the probability of eachnucleotide into full account. A new scoring system has... | Continue reading
We propose a new way of deriving policy gradient updates for reinforcementlearning. Our technique, based on Fourier analysis, recasts integrals thatarise with expected policy gradients as... | Continue reading
In this paper, we propose a drone-based wildfire monitoring system for remoteand hard-to-reach areas. This system utilizes autonomous unmanned aerialvehicles (UAVs) with the main advantage of... | Continue reading
This paper presents a study of semi-supervised learning with largeconvolutional networks. We propose a pipeline, based on a teacher/studentparadigm, that leverages a large collection of... | Continue reading
We show how any dataset of any modality (time-series, images, sound...) canbe approximated by a well-behaved (continuous, differentiable...) scalarfunction with a single real-valued parameter.... | Continue reading
Recent advances in sound engineering have led to the development of so-calledstreamlined music designed to reduce exogenous attention and improve endogenousattention. Although anecdotal reports... | Continue reading
Treating differentials as independent algebraic units have a long history ofuse and abuse. It is generally considered problematic to treat the derivativeas a fraction of differentials rather... | Continue reading
Analyzing the behavior of a program running on a processor that supportsspeculative execution is crucial for applications such as execution timeestimation and side channel detection.... | Continue reading
Alternatives to recurrent neural networks, in particular, architectures basedon attention or convolutions, have been gaining momentum for processing inputsequences. In spite of their relevance,... | Continue reading
Recent work has sought to understand the behavior of neural networks bycomparing representations between layers and between different trained models.We examine methods for comparing neural... | Continue reading
We study selfish mining in Ethereum. The problem is combinatorially morecomplex than in Bitcoin because of major differences in the reward system and adifferent difficulty adjustment formula.... | Continue reading
In this paper, we study multi-armed bandit problems in explore-then-commitsetting. In our proposed explore-then-commit setting, the goal is to identifythe best arm after a pure experimentation... | Continue reading
Understanding human's language requires complex world knowledge. However,existing large-scale knowledge graphs mainly focus on knowledge about entitieswhile ignoring knowledge about activities,... | Continue reading
Behavior and motion planning play an important role in automated driving.Traditionally, behavior planners instruct local motion planners with predefinedbehaviors. Due to the high scene... | Continue reading
The dream of programming language design is to bring aboutorders-of-magnitude productivity improvements in software development tasks.Designers can endlessly debate on how this dream can be... | Continue reading
In the quest to align deep learning with the sciences to address calls forrigor, safety, and interpretability in machine learning systems, thiscontribution identifies key missing pieces: the... | Continue reading
Given an outfit, what small changes would most improve its fashionability?This question presents an intriguing new vision challenge. We introduceFashion++, an approach that proposes minimal... | Continue reading
Dropout methods are a family of stochastic techniques used in neural networktraining or inference that have generated significant research interest and arewidely used in practice. They have been... | Continue reading
Understanding the core content of quantum mechanics requires us todisentangle the hidden logical relationships between the postulates of thistheory. Here we show that the mathematical structure... | Continue reading
Facial pose estimation has gained a lot of attentions in many practicalapplications, such as human-robot interaction, gaze estimation and drivermonitoring. Meanwhile, end-to-end deep... | Continue reading
We present techniques for automatically inferring invariant properties offeed-forward neural networks. Our insight is that feed forward networks shouldbe able to learn a decision logic that is... | Continue reading
Autonomous mechanisms have been proposed to regulate certain aspects ofsociety and are already being used to regulate business organisations. We takeseriously recent proposals for algorithmic... | Continue reading
In order to solve complex, long-horizon tasks, intelligent robots need to beable to carry out high-level, abstract planning and reasoning in conjunctionwith motion planning. However, abstract... | Continue reading
Human-robot interaction often occurs in the form of instructions given from ahuman to a robot. For a robot to successfully follow instructions, a commonrepresentation of the world and objects in... | Continue reading
WebGL is a browser feature that enables JavaScript-based control of thegraphics processing unit (GPU) to render interactive 3D and 2D graphics,without the use of plug-ins. Exploiting WebGL for... | Continue reading
Analog machine learning hardware platforms promise to be faster and moreenergy-efficient than their digital counterparts. Wave physics, as found inacoustics and optics, is a natural candidate... | Continue reading
A new a priori estimate for solutions to Navier-Stokes equations is derived.Uniqueness and existence of these solutions in $\mathbb{R}^3$ for all $t>0$ isproved in a class of solutions locally... | Continue reading
Despite its success, deep learning still needs large labeled datasets tosucceed. Data augmentation has shown much promise in alleviating the need formore labeled data, but it so far has mostly... | Continue reading
Correct concurrent programs are difficult to write; when multiple threadsmutate shared data, they may lose writes, corrupt data, or produce erraticprogram behavior. While many of the data-race... | Continue reading
Rather than proposing a new method, this paper investigates an issue presentin existing learning algorithms. We study the learning dynamics ofreinforcement learning (RL), specifically a... | Continue reading
We propose the Neural Logic Machine (NLM), a neural-symbolic architecture forboth inductive learning and logic reasoning. NLMs exploit the power of bothneural networks---as function... | Continue reading