Transfer Learning for Human & AI

We derive a lumped circuit model for inductive antenna spin-wave transducers in the vicinity of a ferromagnetic medium . The model considers the antenna’sOhmic resistance, its inductance, as well as the additional inductance due to the excitation of resonance or spin waves .…

AciKV is a middle ground between systems that perform fast atomic operations and those that support transactions . This middle ground is useful as it providessimilar performance to the former, while preventing isolation and consistencyanomalies like the latter . Under workloads that involve write requests, AciV exhibits more than two orders of magnitude higher throughput than existing strongly durable systems .…

Neuromorphic (brain-like) processors are one highly-promising solution to the ever-increasing demand for AI-optimised hardware . The photonic system is combined with a software-implemented spiking neural network yielding a full platform for complex image classification tasks . This work highlights potentials of VCSEL-based platforms for novel, ultrafast, all-optical processors interfacing with current computation and communications systems for use in future light-enabled AI and computer vision functionalities .…

Today’s database systems have shown to be capable of supporting AI applications that demand a lot of data processing . Mongolog is syntactically a subset of Prolog language, and we define its operational semantics through translations into aggregation pipelines . Queries in MongoDBdatabases are represented as sequential stages within an aggregation pipelinewhere each stage defines a transformation of the input data, and passes thetransformed data to the next stage .…

The Distance Geometry Problem asks for a realization of a given weightedgraph in $\mathbb{R}^K$ Two variants of this problem, both originating fromprotein conformation, are based on a given vertices . The presence of this constraint allows one to prove, via a combinatorial counting of the number of solutions,that the realization algorithm is fixed-parameter tractable .…

DLBench is a benchmark to evaluate and compare data lake systems that support textual and/or tabular contents . DLBench proposes a data model made of both textual and raw tabular documents, a load model composed of various tasks, as well as a set of performance-based metrics, all relevant to the context of data lakes .…

In the Intersection Non-Emptiness problem, we are given a list of finiteautomata $A_1,A_2,\dots, A_m$ over a common alphabet $W\in \Sigma^*$ as input . We show that the problem is NP-complete when all input automata accept languages of the levels zero or one half .…

The Distance Geometry Problem asks for a realization of a given weightedgraph in $\mathbb{R}^K$ Two variants of this problem, both originating fromprotein conformation, are based on a given vertices . The presence of this constraint allows one to prove, via a combinatorial counting of the number of solutions,that the realization algorithm is fixed-parameter tractable .…

Spatial structures in the 3D space are important to determine molecularproperties . Recent papers use geometric deep learning to represent molecules and predict properties . These papers have not considered thenon-uniformity of interatomic distances . To deal with such issues, we introduce a variant of the Transformer for molecularrepresentations .…

The vision transformer (ViT)achieves promising results on computer vision due to its attention mechanism . However, with the fixed size of patches in ViT, the class token in deeplayer focuses on the global receptive field and cannot generatemulti-granularity features for FGVR .…

Determining the matrix multiplication exponent is one of the most open problems in theoretical computer science . We show that it isimpossible to prove $1_A$ by starting with structure tensors of modules of modulesof fixed degree and using arbitrary restrictions .…

This paper introduces a Python library called pyFFS for efficient FScoefficient computation, convolution, and interpolation . PyFFS addresses the computation of FSco coefficients through what we call the fast Fourier series (FFS) It is available as an open source package athttps://://github.com/imagingofthings/pyFFS,…

The tendency of a jet to stay attached to a flat or convex surface is called the Coand\u{a} effect and has many potential technical applications . The aim of this thesis is to assess how well Computational Fluid Dynamics can capture it .…

Deep learning algorithm for solving Principal-Agent (PA)mean field games with market-clearing conditions . We use an actor-critic approach to optimization, where the agents form a Nash equilibria according to the principal’s penalty function . The inner problem’s Nashequilibria is obtained using a variant of the deep backward stochasticdifferential equation (BSDE) method modified for McKean-Vlasov forward-backwardSDEs .…

LazySets.jl is an easy-to-use, general-purpose and scalable library for computations that mix symbolics and numerics . It is the core library of JuliaReach, a cutting-edge software addressing the fundamental problem of reachability analysis . The library was originally designed for reachability and formal verification, but its scope goes beyond such topics .…

We do not directly compute the short rational generating function for $|\mathcal{P} \cap \mathbb{Z}^n|$ We compute its particular representation inthe form of exponential series that depends on only one variable . Theparametric versions of the above problem are also considered .…

Spatial structures in the 3D space are important to determine molecularproperties . Recent papers use geometric deep learning to represent molecules and predict properties . These papers have not considered thenon-uniformity of interatomic distances . To deal with such issues, we introduce a variant of the Transformer for molecularrepresentations .…

Missing values occur commonly in multidimensional data warehouses . They generate problems of usefulness of data since analysis performed on different dimensions with different dimensions . Therefore, it’s essential to complete these missing values in orderto carry out a better analysis .…

An end-to-end model can be critical for policy-makers to compare their outcomes with past strategies in order to devise better plans for future measures . We observed that some interesting contagion phenomena are directly influenced by mobilityrestrictions and curfew policies .…

Explainable Artificial Intelligence (XAI) is increasingly expandingmore areas of application . Little has been applied to make deep ReinforcementLearning (RL) more comprehensible . This study proposes anovel approach to explain cooperative strategies in multiagent RL using Shapleyvalues . Shapley values are a game theory concept used in XAI that successfully explains therationale behind decisions taken by Machine Learning algorithms .…

A widely-used similarity measure is the graph edit distance, which provides an intuitive notion of similarity . Since its computation is NP-hard, techniques for accelerating similarity search have been studied extensively . We propose a filter-verification framework for similaritysearch, which supports non-uniform edit costs for graphs with arbitrary attributes .…

Spatial structures in the 3D space are important to determine molecularproperties . Recent papers use geometric deep learning to represent molecules and predict properties . These papers have not considered thenon-uniformity of interatomic distances . To deal with such issues, we introduce a variant of the Transformer for molecularrepresentations .…

The tendency of a jet to stay attached to a flat or convex surface is called the Coand\u{a} effect and has many potential technical applications . The aim of this thesis is to assess how well Computational Fluid Dynamics can capture it .…

DLBench is a benchmark to evaluate and compare data lake systems that support textual and/or tabular contents . DLBench proposes a data model made of both textual and raw tabular documents, a load model composed of various tasks, as well as a set of performance-based metrics, all relevant to the context of data lakes .…

In this paper, we present a simple yet effective formulation called CoverageAxis for 3D shape skeletonization . Inspired by the set cover problem, our keyidea is to cover all the surface points using as few inside medial balls as possible .…

The Distance Geometry Problem asks for a realization of a given weightedgraph in $\mathbb{R}^K$ Two variants of this problem, both originating fromprotein conformation, are based on a given vertices . The presence of this constraint allows one to prove, via a combinatorial counting of the number of solutions,that the realization algorithm is fixed-parameter tractable .…

We discuss the notion of a dense cluster with respect to the informationdistance . We prove that all such clusters have an extractable core that represents the mutual information shared by the objects in the cluster .…

In various real-world tasks, an agent needs tocooperate with unknown partner agent types . This requires the agent to assessthe behaviour of the partner agent during a cooperative task and to adjust itsown policy to support the cooperation . Deep reinforcement learning models can be trained to deliver the required functionality but are known to suffer from sample inefficiency and slow learning .…

LazySets.jl is an easy-to-use, general-purpose and scalable library for computations that mix symbolics and numerics . It is the core library of JuliaReach, a cutting-edge software addressing the fundamental problem of reachability analysis . The library was originally designed for reachability and formal verification, but its scope goes beyond such topics .…

The traditional finite-difference time-domain (FDTD) method is constrained by the Courant-Friedrich-Levy (CFL) condition . By considering the partially filled cells, the proposed CLOD-FDTDmethod can significantly improve the accuracy of the FDTD method . At the same time the proposed methodpreserves unconditional stability, which is analyzed and numerically validatedusing the Von-Neuman method .…

We consider zero-sum games on infinite graphs, with objectives specified assets of infinite words over some alphabet of colors . We focus on the strategycomplexity question: given an objective, how much memory does each player need to play as well as possible?…

This paper introduces a Python library called pyFFS for efficient FScoefficient computation, convolution, and interpolation . PyFFS addresses the computation of FSco coefficients through what we call the fast Fourier series (FFS) It is available as an open source package athttps://://github.com/imagingofthings/pyFFS,…

Continuous double auctions are widely used in practice to match buyers and sellers of a variety of offinancial instruments . We develop an agent-based model for trading in a limit order book . We show how opponent modelling techniquescan can be applied to classify trading agent archetypes and (2) how behaviouralcloning can be used to imitate these agents in a simulated setting .…

The graph case of the graph case was proved by Alon, Fischer, Newman andShapira . The error term measuring thequasirandomness of the subhypergraphs of $H$ requires only an arbitrarily smalladditive correction . This has applications to combinatorial property testing . We prove that with high probability a large enoughuniform random sample of the vertices of $K$ also admits the same regularityinstance .…

We do not directly compute the short rational generating function for $|\mathcal{P} \cap \mathbb{Z}^n|$ We compute its particular representation inthe form of exponential series that depends on only one variable . Theparametric versions of the above problem are also considered .…

The graph case of the graph case was proved by Alon, Fischer, Newman andShapira . The error term measuring thequasirandomness of the subhypergraphs of $H$ requires only an arbitrarily smalladditive correction . This has applications to combinatorial property testing . We prove that with high probability a large enoughuniform random sample of the vertices of $K$ also admits the same regularityinstance .…

Intelligent reflecting surface (IRS) has emerged as a key enabling technology to realize smart and reconfigurable radio environment for wireless communications . IRS provides a new and cost-effectivemeans to combat the wireless channel impair-ments in a proactive manner . IRS faces new and unique challenges in its efficient integration into wireless communication systems, especially its efficiency and passive beamforming design under various practical hardware constraints .…

Rust is a general-purpose programming language that is both type- andmemory-safe . Rust achieves theseproperties through a sophisticated, but complex, type system . Wedesigned Bronze, an optional, library-based garbage collector for Rust, makes Rust relatively hard to learn and use .…

Rust is a general-purpose programming language that is both type- andmemory-safe . Rust achieves theseproperties through a sophisticated, but complex, type system . Wedesigned Bronze, an optional, library-based garbage collector for Rust, makes Rust relatively hard to learn and use .…

In this paper, we present a simple yet effective formulation called CoverageAxis for 3D shape skeletonization . Inspired by the set cover problem, our keyidea is to cover all the surface points using as few inside medial balls as possible .…

Over the past 27 years, quantum computing has seen a huge rise in interest from both academia and industry . The importance of verifying programs is especially important in the quantum setting due to how difficult it is to program complex algorithmscorrectly on resource-constrained and error-prone quantum hardware .…

Determining the matrix multiplication exponent is one of the most open problems in theoretical computer science . We show that it isimpossible to prove $1_A$ by starting with structure tensors of modules of modulesof fixed degree and using arbitrary restrictions .…

Circular (or cyclic) proofs have received increasing attention in recent years . Little is known about the complexity theoretic aspects of circular proofs . We introduce a circular proof system based on Bellantoni andCook’s famous safe-normal function algebra . We identify suitable prooftheoretical constraints to characterise the polynomial-time and elementarycomputable functions .…

An end-to-end model can be critical for policy-makers to compare their outcomes with past strategies in order to devise better plans for future measures . We observed that some interesting contagion phenomena are directly influenced by mobilityrestrictions and curfew policies .…

In EHR data, a patient’s history is often represented as a sequence of visits, and each visit contains multiple events . As a result, deeplearning models developed for sequence modeling, like recurrent neural networks, are common architecture for EHR-based clinical events predictive models .…

Automatic speech recognition (ASR) is a capability which enables a program toprocessing human speech into a written form . The current developments in biologically-inspired ASR models, based on spikingneural networks (SNNs), lag behind in terms of accuracy and focus primarily on small scale applications .…

Determining the matrix multiplication exponent is one of the most open problems in theoretical computer science . We show that it isimpossible to prove $1_A$ by starting with structure tensors of modules of modulesof fixed degree and using arbitrary restrictions .…

AciKV is a middle ground between systems that perform fast atomic operations and those that support transactions . This middle ground is useful as it providessimilar performance to the former, while preventing isolation and consistencyanomalies like the latter . Under workloads that involve write requests, AciV exhibits more than two orders of magnitude higher throughput than existing strongly durable systems .…

We consider zero-sum games on infinite graphs, with objectives specified assets of infinite words over some alphabet of colors . We focus on the strategycomplexity question: given an objective, how much memory does each player need to play as well as possible?…

We do not directly compute the short rational generating function for $|\mathcal{P} \cap \mathbb{Z}^n|$ We compute its particular representation inthe form of exponential series that depends on only one variable . Theparametric versions of the above problem are also considered .…