Deep Contextual Video Compression

Most of the existing neural video compression methods adopt the predictivecoding framework, which first generates the predicted frame and then encodesits residue with the current frame . To tap the potential of conditional coding, we proposeusing feature domain context as condition .…

preCICE v2 A Sustainable and User Friendly Coupling Library

preCICE is a free/open-source coupling library . It enables creatingpartitioned multi-physics simulations by gluing together separate software packages . We cover the fundamentals of the software alongside aperformance and accuracy analysis of different data mapping methods . Afterwards, we describe ready-to-use integration with widely-used externalsimulation software packages, tests and continuous integration from unit tosystem level, and community building measures, drawing an overview of thepreCICE ecosystem .…

Process discovery on deviant traces and other stranger things

Process discovery research field has gained more importance, developing two different classes of approaches to modelrepresentation . We focus on declarative processes and embrace the less-popular view of process discovery as a binary supervisedlearning task . We therefore deepen how valuable information brought by both sets can be extracted and formalised into a model that is”optimal” according to user-defined goals .…

Rectangular Spiral Galaxies are Still Hard

Spiral Galaxies is a pencil-and-paper puzzle played on a grid of unitsquares . The goal is to partition the grid into polyominoes such that each center contains exactly one center and is rotationally symmetric about its center . We prove NP-completeness of the designproblem of minimizing the number of centers such that there exist a set ofSpiral Galaxies that exactly cover a given shape .…

Benchmarking a Probabilistic Coprocessor

Computation in the past decades has been driven by deterministic computers based on classical deterministic bits . Recently, alternative computing paradigms like quantum computing and probabilistic computing have gained traction . To efficiently implement Monte Carlo algorithms the generation of random numbers is crucial .…

Approximating the Bundled Crossing Number

There is a polynomial time algorithm to compute an 8-approximation of the bundled crossing number of a good drawing (upto adding a term depending on the facial structure of the drawing) In the special case of circular drawings the approximation factor is 8 (no extraterm), this improves upon the 10-approxyimation of Fink et al.…

Modeling Interactions of Autonomous Vehicles and Pedestrians with Deep Multi Agent Reinforcement Learning for Collision Avoidance

The sequential nature of thevehicle-pedestrian interaction is an often neglected but important aspect of the interaction . We model the corresponding interaction sequence as a Markov decision process (MDP) that is solved by deepreinforcement learning (DRL) algorithms . The results show that the AV is able to mitigate collisions under the majority of the investigated conditions and that the DRL-based pedestrian model indeed learns a more human-like crossing behavior .…

Approximating the Bundled Crossing Number

There is a polynomial time algorithm to compute an 8-approximation of the bundled crossing number of a good drawing (upto adding a term depending on the facial structure of the drawing) In the special case of circular drawings the approximation factor is 8 (no extraterm), this improves upon the 10-approxyimation of Fink et al.…

Rectangular Spiral Galaxies are Still Hard

Spiral Galaxies is a pencil-and-paper puzzle played on a grid of unitsquares . The goal is to partition the grid into polyominoes such that each center contains exactly one center and is rotationally symmetric about its center . We prove NP-completeness of the designproblem of minimizing the number of centers such that there exist a set ofSpiral Galaxies that exactly cover a given shape .…

Unified Shader Programming in C

In real-time graphics, the strict separation of programming languages and environments for host (CPU) code and GPU code results in code duplication, compatibility bugs, and additional development and maintenance costs . Popular general-purpose GPU (GPGPU) programming models like CUDA andC++ AMP avoid many of these issues by presenting unified programming environments .…

Combining Sobolev Smoothing with Parameterized Shape Optimization

Sobolev gradient smoothing can considerably improve the performance of aerodynamic shape optimization and prevent issues with regularity . The incorporation of the parameterizationallows for direct application to engineering test cases, where shapes are described by a CAD model . The new methodology presented in this paper is used for reference test cases from aerodynamics shape optimization, and performance improvements in comparison to a classical Quasi-Newton scheme are shown .…

Width Based Planning and Active Learning for Atari

Width-based planning has shown promising results on Atari 2600 games usingpixel input . Recent width-based approaches have computed featurevectors for each screen using a hand designed feature set or a variationalautoencoder (VAE) trained on game screens . In this paper, we explore consideration ofuncertainty in features generated by a VAE during planning .…

Learning to Superoptimize Real world Programs

SelfImitation Learning for Optimization (SILO) superoptimizes programs an expected 6.2% of our test set . SILO’s rate of superoptimization onour test set is over five times that of a standard policy gradient approach and a model pre-trained on compiler optimization demonstration.…

Neural Knitworks Patched Neural Implicit Representation Networks

Coordinate-based Multilayer Perceptron (MLP) networks are not performant for internal image synthesis applications . Convolutional Neural Networks (CNNs) are typically used instead for a variety of internal generative tasks, at the cost of a larger model . We propose Neural Knitwork, an architecture for neuralimplicit representation learning of natural images that achieves imagesynthesis by optimizing the distribution of image patches in an adversarialmanner .…

Unified Shader Programming in C

In real-time graphics, the strict separation of programming languages and environments for host (CPU) code and GPU code results in code duplication, compatibility bugs, and additional development and maintenance costs . Popular general-purpose GPU (GPGPU) programming models like CUDA andC++ AMP avoid many of these issues by presenting unified programming environments .…

Finite State Mutual Dimension

In 2004, Dai, Lathrop, Lutz, and Mayordomo defined and investigated thefinite-state dimension (a finite-state version of algorithmic dimension) of asequence $S \in \Sigma^\infty$ In 2018, Case and Lutz defined and .investigated the mutual (algorithmic) dimension between two sequences $S . and $T .…

Common Artifacts in Volume Rendering

Direct Volume Rendering is a popular and powerful visualization method for volumetric scalar data sets . In medical applications volume rendering is very commonly used, and has become one of the state of the art methods for 3D visualization of medical data .…

Certified Everlasting Zero Knowledge Proof for QMA

In known constructions of classical zero-knowledge protocols for NP, eitherof zero knowledge or soundness holds only against computationally boundedadversaries . In this work, we introduce a novelcompromise, which we call the certified everlastingzero-knowledge proof for QMA . The verifier issues a classical certificate that shows that the verifier has deleted itsquantum information .…

Power Consumption Analysis of Parallel Algorithms on GPUs

Due to their highly parallel multi-cores architecture, GPUs are being increasingly used in a wide range of computationally intensive applications . Compared to CPUs, they can achieve higher performances at accelerating the programs’ execution in an energy-efficient way . Using our methodology, energy needs of a program could bepredicted, allowing a better task scheduling, according to a new developed energy measurement protocol .…

Audio to Image Cross Modal Generation

Cross-modal representation learning allows to integrate information from different modalities into one representation . Research ongenerative models tends to focus on the visual domain with less emphasis on other domains, such as audio or text . Studies successfully linking more than one modality inthe generative setting are rare .…

A Computable Piece of Uncomputable Art whose Expansion May Explain the Universe in Software Space

At the intersection of what I call uncomputable art and computationalepistemology, we find an exciting and promising area of science related to causation with an alternative, possibly best possible, solution to the challenge of the inverse problem . That is theproblem of finding the possible causes, mechanistic origins, first principles, and generative models of a piece of data from a physical phenomenon .…

Relational Memory Native In Memory Accesses on Rows and Columns

Analytical database systems are typically designed to use a column-first datalayout that maps better to analytical queries of access patterns . We propose a new database management system (DBMS) architecture that is the first hardware/software co-design . It relies on an FPGA-basedaccelerator to transparently transform base data to any group of columns with minimal overhead at runtime at runtime .…

Pursuit evasion games on latin square graphs

We investigate various pursuit-evasion parameters on latin square graphs . The cop number, metric dimension, and localization number are studied . The metric dimension of back-circulant latinsquares shows that the lower bound is close to tight . Recent results on coversand partial transversals of latin squares provide the upper bound of $n+O\left(\frac{\log{n]\right) on the localization number of alatin square graphs of order $n.$…

Visualization Design Sprints for Online and On Campus Courses

A design sprint is a unique process based on rapid prototyping and user testing to define goals and validate ideas before costly development . The well-defined, interactive, and time-constrained design cycle makes design sprints a promising option for teaching project-based and active-learning-centered courses to increase studentengagement and hands-on experience .…

Identity Expression Ambiguity in 3D Morphable Face Models

3D Morphable Models are a class of generative models commonly used to modelfaces . They are typically applied to ill-posed problems such as 3Dreconstruction from 2D data . We demonstrate that non-orthogonality of the variation in variation in identity and expression can causeidentity-expression ambiguity in 3Dmorphable Models .…

Gaps Ambiguity and Establishing Complexity Class Containments via Iterative Constant Setting

Cai and Hemachandra used iterative constant-setting to prove that Few$\subseteq$ $\oplus$P (and thus that FewP $1) is a restricted counting class . Their work lowers the bar for what advances regarding the existence of infinite, P-printable sets of primes would suffice to show that restricted counting classes based on the primes have the power to accept superconstant-ambiguity analogues of UP .…

The Neural MMO Platform for Massively Multiagent Research

Neural MMO is a computationally accessible research platform that combines large agent populations, long time horizons, open-ended tasks, and modular gamesystems . We present Neural MMO as free and opensource software with active support, ongoing development, documentation, and training, logging, and visualization tools to help users adapt to the new setting .…