The origins of the high-energy cosmic neutrino flux remain largely unknown.Recently, one high-energy neutrino was associated with a tidal disruption event(TDE). Here we present AT2019fdr, an exceptionally luminous TDE candidate,coincident with another high-energy neutrino. Our ob … | Continue reading
After the success of the Bitcoin blockchain, came several cryptocurrenciesand blockchain solutions in the last decade. Nonetheless, Blockchain-basedsystems still suffer from low transaction rates and high transaction processinglatencies, which hinder blockchains' scalability. An … | Continue reading
The 2022 Russian invasion of Ukraine is being fought on two fronts: a brutalground war and a duplicitous disinformation campaign designed to conceal andjustify Russia's actions. This campaign includes at least one example of adeep-fake video purportedly showing Ukrainian Presiden … | Continue reading
SHACL is a W3C-proposed language for expressing structural constraints on RDFgraphs. In recent years, SHACL's popularity has risen quickly. This rise inpopularity comes with questions related to its place in the semantic web,particularly about its relation to OWL (the de facto st … | Continue reading
Pretraining on noisy, internet-scale datasets has been heavily studied as atechnique for training models with broad, general capabilities for text,images, and other modalities. However, for many sequential decision domainssuch as robotics, video games, and computer use, publicly … | Continue reading
Long-range e.g. Coulomb-like interactions for (quantized) topological chargesin liquid crystals are observed experimentally, bringing open question thisarticle is focused on: how far can we take this similarity with particlephysics? Uniaxial nematic liquid crystal of ellipsoid-li … | Continue reading
The Bayesian approach to data analysis provides a powerful way to handleuncertainty in all observations, model parameters, and model structure usingprobability theory. Probabilistic programming languages make it easier tospecify and fit Bayesian models, but this still leaves us w … | Continue reading
We present Aeneas, a new verification toolchain for Rust programs based on alightweight functional translation. We leverage Rust's rich region-based typesystem to eliminate memory reasoning for many Rust programs, as long as they donot rely on interior mutability or unsafe code. … | Continue reading
I often work on my laptop in bed. When needed, I park the laptop on thebedside table, where the computer has to share the small available space with alamp, books, notes, and heaven knows what else. It often gets quite squeezy. Being regularly faced with this tricky situation, it … | Continue reading
Many benefits of online communities---such as obtaining new information,opportunities, and social connections---increase with size. Thus, a``successful'' online community often evokes an image of hundreds of thousandsof users, and practitioners and researchers alike have sought t … | Continue reading
We summarise different aspects of the measurement problem in quantummechanics. We argue that it is a real problem which requires a solution, andidentify the properties a theory needs to solve the problem. We show that nocurrent interpretation of quantum mechanics solves the probl … | Continue reading
Understanding to what extent neural networks memorize training data is anintriguing question with practical and theoretical implications. In this paperwe show that in some cases a significant fraction of the training data can infact be reconstructed from the parameters of a train … | Continue reading
Commonsense causality reasoning (CCR) aims at identifying plausible causesand effects in natural language descriptions that are deemed reasonable by anaverage person. Although being of great academic and practical interest, thisproblem is still shadowed by the lack of a well-pose … | Continue reading
We use many search engines on the Internet in our daily lives. However, theyare not perfect. Their scoring function may not model our intent or they mayaccept only text queries even though we want to carry out a similar imagesearch. In such cases, we need to make a compromise: We … | Continue reading
We describe a novel lossy compression approach called DiffC which is based onunconditional diffusion generative models. Unlike modern compression schemeswhich rely on transform coding and quantization to restrict the transmittedinformation, DiffC relies on the efficient communica … | Continue reading
This paper pursues the insight that large language models (LLMs) trained togenerate code can vastly improve the effectiveness of mutation operatorsapplied to programs in genetic programming (GP). Because such LLMs benefit fromtraining data that includes sequential changes and mod … | Continue reading
The Neural Tangent Kernel (NTK), defined as $Θ_θ^f(x_1, x_2) =\left[\partial f(θ, x_1)\big/\partial θ\right] \left[\partialf(θ, x_2)\big/\partial θ\right]^T$ where $\left[\partial f(θ,\cdot)\big/\partial θ\right]$ is a neural network (NN) Jacobian, hasemerged as a central object … | Continue reading
Non-Fungible Tokens (NFTs) have emerged as a way to collect digital art aswell as an investment vehicle. Despite having been popularized only recently,NFT markets have witnessed several high-profile (and high-value) asset salesand a tremendous growth in trading volumes over the l … | Continue reading
We introduce Hyperbard, a dataset of diverse relational data representationsderived from Shakespeare's plays. Our representations range from simple graphscapturing character co-occurrence in single scenes to hypergraphs encodingcomplex communication settings and character contrib … | Continue reading
Virtual Reality (VR) provides new possibilities for modern knowledge work.However, the potential advantages of virtual work environments can only be usedif it is feasible to work in them for an extended period of time. Until now,there are limited studies of long-term effects when … | Continue reading
Diffusion probabilistic models (DPMs) have become a popular approach toconditional generation, due to their promising results and support forcross-modal synthesis. A key desideratum in conditional synthesis is to achievehigh correspondence between the conditioning input and gener … | Continue reading
Even the largest neural networks make errors, and once-correct predictionscan become invalid as the world changes. Model editors make local updates tothe behavior of base (pre-trained) models to inject updated knowledge orcorrect undesirable behaviors. Existing model editors have … | Continue reading
This book uses the modern theory of artificial intelligence (AI) tounderstand human suffering or mental pain. Both humans and sophisticated AIagents process information about the world in order to achieve goals and obtainrewards, which is why AI can be used as a model of the huma … | Continue reading
State-of-the-art 3D-aware generative models rely on coordinate-based MLPs toparameterize 3D radiance fields. While demonstrating impressive results,querying an MLP for every sample along each ray leads to slow rendering.Therefore, existing approaches often render low-resolution f … | Continue reading
Human vision possesses a special type of visual processing systems calledperipheral vision. Partitioning the entire visual field into multiple contourregions based on the distance to the center of our gaze, the peripheral visionprovides us the ability to perceive various visual f … | Continue reading
Large language models (LMs) offer unprecedented language generationcapabilities and exciting opportunities for interaction design. However, theirhighly context-dependent capabilities are difficult to grasp and are oftensubjectively interpreted. In this paper, we argue that by cur … | Continue reading
We present a modern C++17-compatible thread pool implementation, built fromscratch with high-performance scientific computing in mind. The thread pool isimplemented as a single lightweight and self-contained class, and does not haveany dependencies other than the C++17 standard l … | Continue reading
More than ten years ago the blockchain was acclaimed as the solution toovercome centralised trusted third parties for online payments. Through theyears the crypto-movement changed and evolved, although decentralisationremained the core ideology and the necessary feature every new … | Continue reading
Score-based diffusion models are a class of generative models whose dynamicsis described by stochastic differential equations that map noise into data.While recent works have started to lay down a theoretical foundation for thesemodels, an analytical understanding of the role of … | Continue reading
We study the use of amortized optimization to predict optimal transport (OT)maps from the input measures, which we call Meta OT. This helps repeatedlysolve similar OT problems between different measures by leveraging theknowledge and information present from past problems to rapi … | Continue reading
We empirically analyze five online communities: Friendster, Livejournal,Facebook, Orkut, Myspace, to identify causes for the decline of socialnetworks. We define social resilience as the ability of a community towithstand changes. We do not argue about the cause of such changes, … | Continue reading
Machine knitted textiles are complex multi-scale material structuresincreasingly important in many industries, including consumer products,architecture, composites, medical, and military. Computational modeling,simulation, and design of industrial fabrics require efficient repres … | Continue reading
Network architecture plays a key role in the deep learning-based computervision system. The widely-used convolutional neural network and transformertreat the image as a grid or sequence structure, which is not flexible tocapture irregular and complex objects. In this paper, we pr … | Continue reading
Machine learning (ML) and artificial intelligence (AI) tools increasinglypermeate every possible social, political, and economic sphere; sorting,taxonomizing and predicting complex human behaviour and social phenomena.However, from fallacious and naive groundings regarding comple … | Continue reading
The Hypersyn protocol is a new type of permissionless and peer-to-peerpayment network that is based on the concept of mutual credit and mutualarbitrage. Unlike blockchain-based systems, Hypersyn does not rely on anyconsensus algorithm. It does not require a distributed ledger to … | Continue reading
Technologies for abusive language detection are being developed and appliedwith little consideration of their potential biases. We examine racial bias infive different sets of Twitter data annotated for hate speech and abusivelanguage. We train classifiers on these datasets and c … | Continue reading
Bitcoin is a digital currency designed to rely on a decentralized, trustlessnetwork of anonymous agents. Using a pseudonymous-address-linking procedurethat achieves >99% sensitivity and >99% specificity, we reveal that betweenlaunch (January 3rd, 2009), and when the price reached … | Continue reading
Training foundation models, such as GPT-3 and PaLM, can be extremelyexpensive, often involving tens of thousands of GPUs running continuously formonths. These models are typically trained in specialized clusters featuringfast, homogeneous interconnects and using carefully designe … | Continue reading
Electrical control of superconductivity is critical for nanoscalesuperconducting circuits including cryogenic memory elements, superconductingfield-effect transistors (FETs), and gate-tunable qubits. Superconducting FETsoperate through continuous tuning of carrier density, but th … | Continue reading
Recent work in cognitive science has uncovered a diversity of explanatoryvalues, or dimensions along which we judge explanations as better or worse. Wepropose a Bayesian account of how these values fit together to guideexplanation. The resulting taxonomy provides a set of predict … | Continue reading
Deep learning (DL) has recently changed the development of intelligentsystems and is widely adopted in many real-life applications. Despite theirvarious benefits and potentials, there is a high demand for DL processing indifferent computationally limited and energy-constrained de … | Continue reading
We offer a general theoretical framework for brain and behavior that isevolutionarily and computationally plausible. The brain in our abstract modelis a network of nodes and edges. Although it has some similarities to standardneural network models, as we show, there are some sign … | Continue reading
Recent advances in differentiable rendering have enabled high-qualityreconstruction of 3D scenes from multi-view images. Most methods rely on simplerendering algorithms: pre-filtered direct lighting or learned representationsof irradiance. We show that a more realistic shading mo … | Continue reading
We present an implicit neural representation to learn the spatio-temporalspace of kinematic motions. Unlike previous work that represents motion asdiscrete sequential samples, we propose to express the vast motion space as acontinuous function over time, hence the name Neural Mot … | Continue reading
Large language models such as GPT-3 and PaLM have shown remarkableperformance in few-shot learning. However, they still struggle with reasoningtasks such as the arithmetic benchmark GSM8K. Recent advances deliberatelyguide the language model to generate a chain of reasoning steps … | Continue reading
Image restoration and enhancement is a process of improving the image qualityby removing degradations, such as noise, blur, and resolution degradation. Deeplearning (DL) has recently been applied to image restoration and enhancement.Due to its ill-posed property, plenty of works … | Continue reading