Transfer Learning for Human & AI

In this work, we prove a hypercontractive inequality for matrix-valuedfunctions defined over large alphabets . We present a lower bound for LDCs over largealphabets and present upper and lower bounds for the communication complexity of the Hidden Hypermatching problem . We also show that any streaming algorithm achieving a $(r-\varepsilon)-approximation requires $Omega(n^{1-1/t)$ classical or $Oma(n=2) when $r=2$ is $O(\log{n)$ .…

In this work, we prove a hypercontractive inequality for matrix-valuedfunctions defined over large alphabets . We present a lower bound for LDCs over largealphabets and present upper and lower bounds for the communication complexity of the Hidden Hypermatching problem . We also show that any streaming algorithm achieving a $(r-\varepsilon)-approximation requires $Omega(n^{1-1/t)$ classical or $Oma(n=2) when $r=2$ is $O(\log{n)$ .…

This paper introduces the full Low-carbon Expansion Generation Optimization(LEGO) model available on Github . LEGO is amixed-integer quadratically constrained optimization problem and has beendesigned to be a multi-purpose tool, like a Swiss army knife, that can be used to study many different aspects of the energy sector .…

In this note we present a simplified analysis of the quantum and classicalcomplexity of the $k$-XOR Forrelation problem . We use a stochastic interpretation of the Forrelationdistribution .…

For general input automata, there exist regular constraint languages such that asking if a given input automaton admits a synchronizing word in the constraint language is PSPACE-complete or NP-complete . Here, we investigate the problem for commutative automata over an arbitrary alphabet and automatawith simple idempotents over a binary alphabet .…

We are interested in embedding trees T with maximum degree at most four in arectangular grid . The aim is to determine a straight model of a given treeT minimizing the maximum number of bends over all paths of T .…

In this note we present a simplified analysis of the quantum and classicalcomplexity of the $k$-XOR Forrelation problem . We use a stochastic interpretation of the Forrelationdistribution .…

For general input automata, there exist regular constraint languages such that asking if a given input automaton admits a synchronizing word in the constraint language is PSPACE-complete or NP-complete . Here, we investigate the problem for commutative automata over an arbitrary alphabet and automatawith simple idempotents over a binary alphabet .…

We construct a family of 17 disjoint axis-parallel line segments in the planethat do not admit a circumscribing polygon .…

In this work, we prove a hypercontractive inequality for matrix-valuedfunctions defined over large alphabets . We present a lower bound for LDCs over largealphabets and present upper and lower bounds for the communication complexity of the Hidden Hypermatching problem . We also show that any streaming algorithm achieving a $(r-\varepsilon)-approximation requires $Omega(n^{1-1/t)$ classical or $Oma(n=2) when $r=2$ is $O(\log{n)$ .…

For general input automata, there exist regular constraint languages such that asking if a given input automaton admits a synchronizing word in the constraint language is PSPACE-complete or NP-complete . Here, we investigate the problem for commutative automata over an arbitrary alphabet and automatawith simple idempotents over a binary alphabet .…

Surrogate models have several uses in engineering design, including speedingup design optimization, noise reduction, test measurement interpolation, gradient estimation, portability, and protection of intellectual property . The GSM can accurately predictdisplacement fields from static loads given the structure’s geometry as input, enabling training across multiple parametrizations .…

We are interested in embedding trees T with maximum degree at most four in arectangular grid . The aim is to determine a straight model of a given treeT minimizing the maximum number of bends over all paths of T .…

Estonian Internet voting has been used in national-wide elections since 2005 . The system was initially designed in a heuristic manner, with very few security guarantees . To date, no formal security analysis of the system has been given . For the first time, we provide a rigorous security modeling for the Estonian IVXV system as a ceremony, attempting to capture the effect of actual human behavior on election verifiability in the UC framework .…

Deciding whether a family of disjoint axis-parallel line segments in theplane can be linked into a simple polygon is NP-hard .…

Deciding whether a family of disjoint axis-parallel line segments in theplane can be linked into a simple polygon is NP-hard .…

Opacity is a property expressing whether a system may reveal its secret to an intruder . We design a new algorithm deciding K-step opacity the complexity of which is lower than that of existing algorithms and that does not depend on K .…

Deciding whether a family of disjoint axis-parallel line segments in theplane can be linked into a simple polygon is NP-hard .…

Deciding whether a family of disjoint axis-parallel line segments in theplane can be linked into a simple polygon is NP-hard .…

Deciding whether a family of disjoint axis-parallel line segments in theplane can be linked into a simple polygon is NP-hard .…

For some Maltsev conditions it is enough to check if a finitealgebra satisfies $Sigma$ locally on subsets of bounded size . Thislocal-global property is the main known source of tractability results for deciding Maltsevs conditions . In this paper we investigate the local-globalproperty for the existence of a $G-term, i.e.…

For some Maltsev conditions it is enough to check if a finitealgebra satisfies $Sigma$ locally on subsets of bounded size . Thislocal-global property is the main known source of tractability results for deciding Maltsevs conditions . In this paper we investigate the local-globalproperty for the existence of a $G-term, i.e.…

For some Maltsev conditions it is enough to check if a finitealgebra satisfies $Sigma$ locally on subsets of bounded size . Thislocal-global property is the main known source of tractability results for deciding Maltsevs conditions . In this paper we investigate the local-globalproperty for the existence of a $G-term, i.e.…

Reachability, distance, and matching are some of the most fundamental graph problems that have been of particular interest in dynamic complexity theory . Reachability can be maintained withfirst-order update formulas, or equivalently in DynFO in general graphs with nnodes . We extend the meta-theorem forreachability to distance and bipartite maximum matching with the same bounds .…

Reachability, distance, and matching are some of the most fundamental graph problems that have been of particular interest in dynamic complexity theory . Reachability can be maintained withfirst-order update formulas, or equivalently in DynFO in general graphs with nnodes . We extend the meta-theorem forreachability to distance and bipartite maximum matching with the same bounds .…

Reachability, distance, and matching are some of the most fundamental graph problems that have been of particular interest in dynamic complexity theory . Reachability can be maintained withfirst-order update formulas, or equivalently in DynFO in general graphs with nnodes . We extend the meta-theorem forreachability to distance and bipartite maximum matching with the same bounds .…

A novel addition to these is a technique that uses alignmentpredicates to align traces of the two programs, in order to construct a programalignment automaton . Being guided by predicates is not just beneficial indealing with syntactic dissimilarities, but also in staying relevant to the property .…

The minimum turncost 3d printing problem is NP-hard, even when the region is a simple polygon . We explore efficient optimization of toolpaths based on multiple criteria for large instances of printing problems . For the Buddha, our framework buildstoolpaths over a total of 799,716 nodes across 169 layers, and for the Bunny it buildsstool paths over 812,733 nodes across 360 layers .…

Stochastic Games (SGs) lay the foundation for the study of multi-agentreinforcement learning (MARL) and sequential agent interactions . We derive that computing an approximate Markov Perfect Equilibrium (MPE) in afinite-state discounted Stochastics Game within the exponential precision is\textbf{PPAD}-complete . Thecompleteness result follows the reduction of the fixed-point problem to {\scEnd of the Line}.…

The minimum turncost 3d printing problem is NP-hard, even when the region is a simple polygon . We explore efficient optimization of toolpaths based on multiple criteria for large instances of printing problems . For the Buddha, our framework buildstoolpaths over a total of 799,716 nodes across 169 layers, and for the Bunny it buildsstool paths over 812,733 nodes across 360 layers .…

Length scale control is imposed in topology optimization (TO) to make designsamenable to manufacturing and other functional requirements . The proposed method does not involve additional constraints, and the sensitivitycomputations are automated by expressing the computations in an end-enddifferentiable fashion using the neural net’s library .…

Stochastic Games (SGs) lay the foundation for the study of multi-agentreinforcement learning (MARL) and sequential agent interactions . We derive that computing an approximate Markov Perfect Equilibrium (MPE) in afinite-state discounted Stochastics Game within the exponential precision is\textbf{PPAD}-complete . Thecompleteness result follows the reduction of the fixed-point problem to {\scEnd of the Line}.…

Stochastic Games (SGs) lay the foundation for the study of multi-agentreinforcement learning (MARL) and sequential agent interactions . We derive that computing an approximate Markov Perfect Equilibrium (MPE) in afinite-state discounted Stochastics Game within the exponential precision is\textbf{PPAD}-complete . Thecompleteness result follows the reduction of the fixed-point problem to {\scEnd of the Line}.…

Length scale control is imposed in topology optimization (TO) to make designsamenable to manufacturing and other functional requirements . The proposed method does not involve additional constraints, and the sensitivitycomputations are automated by expressing the computations in an end-enddifferentiable fashion using the neural net’s library .…

In this paper, we propose and analyze a first-order and a second-ordertime-stepping schemes for the anisotropic phase-field dendritic crystal growth model . The proposed schemes are based on an auxiliary variable approach for theAllen-Cahn equation . The idea of the former is to introducesuitable auxiliary variables to facilitate construction of high order stableschemes for a large class of gradient flows .…

Modern GPU datacenters are critical for delivering Deep Learning (DL) models and services in both the research community and industry . When operating adatacenter, optimization of resource scheduling and management can brings significant financial benefits . Achieving this goal requires a deepunderstanding of the job features and user behaviors .…

Support for lower precision computation is becoming more common inaccelerator hardware due to lower power usage, reduced data movement and increased computational performance . However, computational science and engineering (CSE) problems require double precision accuracy in several domains . We seek the best methods for incorporating multiple precisions into the GMRES linear solver; these includeiterative refinement and parallelizable preconditioners .…

This work presents an algorithm for the computation and visualization of an underlying unknown surface from a given net of geodesics . It is based on atheoretical result by the author regarding minimal Gaussian curvature surfaces with geodesic boundary conditions .…

The domain shift between the source and target domain is the main challenge in Cross-Domain Few-Shot Learning . Self-Taught (ST)approach alleviates problem of non-target guidance by constructing task-oriented metric spaces . It helps to transfer source knowledge onto the target tasksand focus on discriminative regions .…

A high order finite difference method is proposed for unstructured meshes to simulate compressible inviscid/viscous flows with/without discontinuities . In this method, the divergence of the flux oneach vertices is computed directly from fluxes nearby by means of high orderleast-square .…

Current recommendation approaches help online merchants predict, for eachvisiting user, which subset of their existing products is the most relevant . We argue that existing recommendation models cannot be used to predict the optimal combination of features that will makenew products serve better the needs of the target audience .…

Federated Learning (FL) makes a large amount of edge computing devices jointly learn a global model without data sharing . In FL, dataare generated in a decentralized manner with high heterogeneity . We analyze the so-called Local SGD, a multi-round estimation procedure that uses intermittent communication to improve communication efficiency .…

Global vaccination rollout effort suffers from vaccine distribution inequality and vaccine acceptance, leading to insufficient public immunity provided by the vaccine products . Findings might help solve thevaccine hesitancy problem by focusing more on individuals’ opinions and behavior, authors say .…

Clustering analysis discovers hidden structures in a data set by partitioning them into disjoint clusters . Common problems of current clustering accuracy measures include overlooking unmatched clusters, biases towards excessive clusters, unstable baselines, and difficult interpretation . J-score quantifies how well the hypothetical clusters produced by clustering analysis recover the trueclasses.…

Semantic information provides intra-class consistency and inter-classdiscriminability beyond visual concepts . However, semantic information is onlyavailable for labeled samples but absent for unlabeled samples . Therefore, it is inevitable to bring a cross-modal bias betweensemantic-guided samples and nonsemantic samples, which results in aninformation asymmetry problem .…

We consider inverse source problems with multi-frequency sparse nearfield measurements . In contrast to the existing near field operator based on the integral over the space variable . We introduce a multi frequency sampling method to reconstruct the source support .…

Cities worldwide struggle with overloaded transportation systems and theirexternalities, such as traffic congestion and emissions . The emergingautonomous transportation technology has a potential to alleviate these issues . But decisions of profit-maximizing operators running large autonomousfleets could have a negative impact on other stakeholders, e.g.,…

User modeling is critical for personalized web applications . Existing usermodeling methods usually train user models from user behaviors with task-specific labeled data . But labeled data in a target task may be insufficient for training accurate user models . Pre-training user models on unlabeled user behavior data has the potential to improve user modeling for many downstream tasks .…

Behavior trees represent a hierarchical and modular way of combining severallow-level control policies into a high-level task-switching policy . Hybriddynamical systems can also be seen in terms of task switching between different policies . A formal continuous-time formulation of behavior trees has been lacking .…

In this paper, we introduce an angle notion, called the singular angle, forstable nonlinear systems from an input-output perspective . The proposed systemsingular angle describes an upper bound for the “rotating effect” from the system input to output . It is, thus, different from the recently appeared nonlinear systemphase which adopts the complexification of real-valued signals using theHilbert transform .…