The "significance filter" refers to focusing exclusively on statisticallysignificant results. Since frequentist properties such as unbiasedness andcoverage are valid only before the data have... | Continue reading
Designing reward functions is difficult: the designer has to specify what todo (what it means to complete the task) as well as what not to do (side effectsthat should be avoided while completing... | Continue reading
From massive face-recognition-based surveillance and machine-learning-baseddecision systems predicting crime recidivism rates, to the move towardsautomated health diagnostic systems, artificial... | Continue reading
It is shown that the set of decimal palindromes is an additive basis for thenatural numbers. Specifically, we prove that every natural number can beexpressed as the sum of forty-nine (possibly... | Continue reading
Deep learning's successes are often attributed to its ability toautomatically discover new representations of the data, rather than relying onhandcrafted features like other learning methods. We... | Continue reading
Despite the advancements in software testing, bugs still plague deployedsoftware and result in crashes in production. When debugging issues --sometimes caused by "heisenbugs" -- there is the... | Continue reading
DNS is important in nearly all interactions on the Internet. All large DNSoperators use IP anycast, announcing servers in BGP from multiple physicallocations to reduce client latency and provide... | Continue reading
Package managers have become a vital part of the modern software developmentprocess. They allow developers to reuse third-party code, share their own code,minimize their codebase, and simplify... | Continue reading
The COVID-19 pandemic has caused a global shortage of personal protectiveequipment. While existing supply chains are struggling to meet the surge indemand, the limited supply of N95 filtering... | Continue reading
Coding standards and good practices are fundamental to a disciplined approachto software projects, whatever programming languages they employ. Prologprogramming can benefit from such an... | Continue reading
We propose a framework for learning neural scene representations directlyfrom images, without 3D supervision. Our key insight is that 3D structure canbe imposed by ensuring that the learned... | Continue reading
As fuzz testing has passed its 30th anniversary, and in the face of theincredible progress in fuzz testing techniques and tools, the question arisesif the classic, basic fuzz technique is still... | Continue reading
Sparse-dense linear algebra is crucial in many domains, but challenging tohandle efficiently on CPUs, GPUs, and accelerators alike; multiplications withsparse formats like CSR and CSF require... | Continue reading
This is an introduction to geometric algebra, an alternative to traditionalvector algebra that expands on it in two ways: 1. In addition to scalars and vectors, it defines new objects... | Continue reading
High performance large scale graph analytics is essential to timely analyzerelationships in big data sets. Conventional processor architectures sufferfrom inefficient resource usage and bad... | Continue reading
In France, income tax is computed from taxpayers' individual returns, usingan algorithm that is authored, designed and maintained by the French PublicFinances Directorate (DGFiP). This algorithm... | Continue reading
The eternal inflation scenario predicts that our observable universe residesinside a single bubble embedded in a vast inflating multiverse. We present thefirst observational tests of eternal... | Continue reading
The music signal comprises of different features like rhythm, timbre, melody,harmony. Its impact on the human brain has been an exciting research topic forthe past several decades.... | Continue reading
The problems in variation here concerned are such as to admit a continuousgroup (in Lie's sense); the conclusions that emerge from the correspondingdifferential equations find their most general... | Continue reading
The rapid progress in artificial intelligence (AI) and machine learning hasopened unprecedented analytics possibilities in various team and individualsports, including baseball, basketball, and... | Continue reading
A number of techniques have been proposed to explain a machine learningmodel's prediction by attributing it to the corresponding input features.Popular among these are techniques that apply the... | Continue reading
Why not have a computer just draw a map? This is something you hear a lotwhen people talk about gerrymandering, and it's easy to think at first thatthis could solve redistricting altogether. But... | Continue reading
We derive the classical Hamilton-Jacobi equation from first principles as thenatural description for smooth stochastic processes when one neglectsstochastic velocity fluctuations. The... | Continue reading
FPGA accelerators are gaining increasing attention in both cloud and edgecomputing because of their hardware flexibility, high computational throughput,and low power consumption. However, the... | Continue reading
In recent years, machine learning has received increased interest both as anacademic research field and as a solution for real-world business problems.However, the deployment of machine learning... | Continue reading
Abstract reasoning and logic inference are difficult problems for neuralnetworks, yet essential to their applicability in highly structured domains. Inthis work we demonstrate that a well known... | Continue reading
We introduce screen gleaning, a TEMPEST attack in which the screen of amobile device is read without a visual line of sight, revealing sensitiveinformation displayed on the phone screen. The... | Continue reading
Gradient descent and backpropagation have enabled neural networks to achieveremarkable results in many real-world applications. Despite ongoing success,training a neural network with gradient... | Continue reading
The OAuth 2.0 protocol is one of the most widely deployedauthorization/single sign-on (SSO) protocols and also serves as the foundationfor the new SSO standard OpenID Connect. Despite the... | Continue reading
Iverson's APL and its descendants (such as J, K and FISh) are examples of thefamily of "rank-polymorphic" programming languages. The principal controlmechanism of such languages is the general... | Continue reading
Most existing datasets for sound event recognition (SER) are relatively smalland/or domain-specific, with the exception of AudioSet, based on a massiveamount of audio tracks from YouTube videos... | Continue reading
Running machine learning analytics over geographically distributed datasetsis a rapidly arising problem in the world of data management policies ensuringprivacy and data security. Visualizing... | Continue reading
Ownership is the concept of tracking aliases and mutations to data, usefulfor both memory safety and system design. The Rust programming languageimplements ownership via the borrow checker, a... | Continue reading
The path toward realizing the potential of seasonal forecasting and itssocioeconomic benefits depends heavily on improving general circulation modelbased dynamical forecasting systems. To... | Continue reading
We examine how the behavior of software developers changes in response toremoving gamification elements from GitHub, an online platform forcollaborative programming and software development. We... | Continue reading
Machine learning has shown growing success in recent years. However, currentmachine learning systems are highly specialized, trained for particularproblems or domains, and typically on a single... | Continue reading
Graph-structured data are an integral part of many application domains,including chemoinformatics, computational biology, neuroimaging, and socialnetwork analysis. Over the last two decades,... | Continue reading
When data is plentiful, the loss achieved by well-trained neural networksscales as a power-law $L \propto N^{-α}$ in the number of networkparameters $N$. This empirical scaling law holds... | Continue reading
Sentiment analysis as a field has come a long way since it was firstintroduced as a task nearly 20 years ago. It has widespread commercialapplications in various domains like marketing, risk... | Continue reading
Code completion is a popular software development tool integrated into allmajor IDEs. Many neural language models have achieved promising results incompletion suggestion prediction on synthetic... | Continue reading
The Appalachian Trail (AT) is a 2193-mile-long hiking trail in the easternUnited States. The trail has many bends and turns at different length scales,which gives it a nontrivial fractal... | Continue reading
We describe a system for large-scale audiovisual translation and dubbing,which translates videos from one language to another. The source language'sspeech content is transcribed to text,... | Continue reading
EfficientNets are a family of state-of-the-art image classification modelsbased on efficiently scaled convolutional neural networks. Currently,EfficientNets can take on the order of days to... | Continue reading
ML models often exhibit unexpectedly poor behavior when they are deployed inreal-world domains. We identify underspecification as a key reason for thesefailures. An ML pipeline is underspecified... | Continue reading
Dask is a distributed task framework which is commonly used by datascientists to parallelize Python code on computing clusters with littleprogramming effort. It uses a sophisticated... | Continue reading
Encrypted database systems provide a great method for protecting sensitivedata in untrusted infrastructures. These systems are built using eitherspecial-purpose cryptographic algorithms that... | Continue reading
The main source of information regarding ancient Mesopotamian history andculture are clay cuneiform tablets. Despite being an invaluable resource, manytablets are fragmented leading to missing... | Continue reading