Programmers will always have a dilemma: they must make a choice between Performance, Flexibility or Fast Development. In this work, an experimental study is presented regarding the advantages of using Assembly languag...
详细信息
ISBN:
(数字)9798331532833
ISBN:
(纸本)9798331532840
Programmers will always have a dilemma: they must make a choice between Performance, Flexibility or Fast Development. In this work, an experimental study is presented regarding the advantages of using Assembly language compared to C and Rust languages from runtime, flexibility and development effort in the cases of two machine learning algorithms, linear regression and k-nearest neighbors, using public datasets. Experiments have shown that for simple ML algorithms, assembly language is an alternative that must be considered. Conclusions as well as future research directions are also included.
Pedestrian trajectory prediction in real-world scenarios is a challenging task for several computer vision applications, such as autonomous driving, video surveillance, and robotic systems. this is not a trivial task ...
详细信息
Pedestrian trajectory prediction in real-world scenarios is a challenging task for several computer vision applications, such as autonomous driving, video surveillance, and robotic systems. this is not a trivial task due to the numerous potential trajectories. In this article, it provides a tree-based approach to handle this multimodal prediction challenge. the tree is designed based on the observed data and is also used to predict future trajectories. In particular, an individual's potential future trajectory is represented by the tree's root-to-leaf route. Compared to previous approaches that use implicit latent variables to describe possible future paths, the movement behaviors may be directly represented by the path in the tree (e.g., go straight and then turn left), and thus offer more socially suitable trajectories. the experimental results on the Eth, UCY, and Stanford Drone datasets show that this approach can exceed the performance of the state-of-the-art approaches. the solution is more efficient and compact, with a smaller model size and a higher accuracy, and delivers better results with reference to average displacement error (ADE) and final displacement error (FDE) metrics.
We investigate in this paper congestion control algorithmsthat are TCP-friendly and media-friendly at the same time, so they are more suitable for multimedia streaming than TCP's AIMD. We use in this investigatio...
详细信息
We investigate in this paper congestion control algorithmsthat are TCP-friendly and media-friendly at the same time, so they are more suitable for multimedia streaming than TCP's AIMD. We use in this investigation the optimization framework developed by Kelly et al. in. Kelly et al. derived two gradient-like algorithms, a primal algorithm and a dual one, to control the congestion in a computer network in an end-to-end way and to obtain optimal bandwidth allocation among competing sources. Starting with Kelly et. al's paper, several other papers proposed different variants of the original primal and dual algorithms, but none of these considers the specific characteristics of multimedia streams. the aim of this paper is to build window-based, primal congestion control algorithms for optimizing the perceived quality of multimedia streams in best-effort networks. We build these media-friendly primal congestion control algorithms using the following recipe: 1) first we construct a window-based congestion control algorithm that is TCP-friendly, but has a much more stable throughput and then 2) we shape this stable throughput according to media characteristics (e.g. bitrate). We present two such primal congestion control algorithms and prove their stability around the equilibrium.
the symbolic-numeric computation has been extensively developed in scientificcomputing for experimenting mathematics in numerical programs, like in optimization problems and finite element methods. Many software and ...
详细信息
ISBN:
(纸本)9781479989386
the symbolic-numeric computation has been extensively developed in scientificcomputing for experimenting mathematics in numerical programs, like in optimization problems and finite element methods. Many software and libraries have been developed to support symbolic-numeric computation especially in the recent years. However, most of the implementations are cumbersome and inefficient for numerically evaluating symbolic expressions. the popular implementation chooses the way that generates C/C++/FORTRAN source codes for symbolic expressions and compiles the source files using the external compilers. the compiled machine codes are then linked back to the symbolic manipulation language environment. thi sprocess suffers from slow compilation and significant overhead of external function calls. To address this problem, this paper presents a handy approach that provides fast numerical evaluation for symbolic expressions in Java. In our approach, Java bytecode is generated in memory for symbolic expressions and further Just-In-Time (JIT) compiled to machine codes onJava Virtual Machine (JVM) at runtime. We have developedSymJava (https://***/yuemingl/SymJava) to implement our approach and tested a range of benchmark problems. the results show that SymJava is 1~3 orders of magnitude faster than the existing implementations including Matlab, Mathematica, Sage, theano and SymPy. Additionally, SymJava offers a human friendly programming style for symbolic expressions by overloading operators in Java. Our approach opens up a new avenue for the development of next generation symbolic-numeric software.
Quantum key distribution (QKD) has seen an important development for the last ten years. Nowadays, different companies offer commercial solutions and key exchanges at pulse rates of up to 1 GHz and over distances of u...
详细信息
Quantum key distribution (QKD) has seen an important development for the last ten years. Nowadays, different companies offer commercial solutions and key exchanges at pulse rates of up to 1 GHz and over distances of up to 200 km have been demonstrated in research labs. However, the generation of random bit strings has often been neglected and it remains a technical problem, as physical random number generators have still insufficient bit rates for high speed QKD. In this talk, after an introduction in QKD, a state of the art prototype developed at the University of Geneva was presented. this prototype allows generating keys at high rates over up to 200 km of optical fiber. However, due to the lack of high rate physical random generators, random bits at a rate of 350 MHz are produced by a pseudo random number generator seeded by random bits originating from a quantum random number generator (4 MHz, id Quantique). It was show how an eavesdropper could attack a QKD system based on pseudo random numbers. Finally, different possibilities to increase the rates of quantum random number generators was discussed.
this paper considers the problem of scheduling jobs on single and parallel machines where all the jobs possess different processing times but a common due date. there is a penalty involved with each job if it is proce...
详细信息
this paper considers the problem of scheduling jobs on single and parallel machines where all the jobs possess different processing times but a common due date. there is a penalty involved with each job if it is processed earlier or later than the due date. the objective of the problem is to find the assignment of jobs to machines, the processing sequence of jobs and the time at which they are processed, which minimizes the total penalty incurred due to tardiness or earliness of the jobs. this work presents exact polynomial algorithms for optimizing a given job sequence for single and parallel machines withthe run-time complexities of O(n log n) and O(mn 2 log n) respectively, where n is the number of jobs and m the number of machines. the algorithms take a sequence consisting of all the jobs (J i , i = 1,2,...,n) as input and distribute the jobs to machines (for m>1) along withtheir best completion times so as to get the least possible total penalty for this sequence. We prove the optimality for the single machine case and the runtime complexities of both. Henceforth, we present the results for the benchmark problems and compare with previous work for both single and parallel machine cases, up to 200 jobs.
Ant systems normally rely on identical agents that cooperate indirectly using pheromone trails to find a problem solution. this uniformity may be inadequate in solving difficult problems where complex behavior pattern...
详细信息
Ant systems normally rely on identical agents that cooperate indirectly using pheromone trails to find a problem solution. this uniformity may be inadequate in solving difficult problems where complex behavior patterns are needed. A new ant-based model is developed withthe aim of promoting non-uniformity in the agent population by enhancing each agent with properties that induce heterogeneity. Agents are endowed with different pheromone sensitivity levels. Highly-sensitive agents are essentially influenced in the decision making process by stigmergic information and thus likely to select strong pheromone-marked moves. Agents with low sensitivity are biased towards random search inducing diversity for exploration. Sensitive agents allow many types of reactions to a changing environment facilitating an efficient balance between exploration and exploitation. this balance is pursued by organizing the agents in several subpopulations based on a sensitivity level interval. therefore, each subpopulation has a specific behavior given by the sensitivity level spanning from the diversification to the intensification of the search. the proposed model is engaged in a set of numerical experiments and comparisons for solving various combinatorial optimization problems with very promising results.
We continue developing the project on working with rational orthogonal matrices. Previous work presented at SYNASC is improved by initiating a database of matrices. this requires two developments: A systematic method ...
详细信息
ISBN:
(数字)9798331532833
ISBN:
(纸本)9798331532840
We continue developing the project on working with rational orthogonal matrices. Previous work presented at SYNASC is improved by initiating a database of matrices. this requires two developments: A systematic method for generating matrices, and a method for organising and presenting the matrices in the database accessible to users. 1
Cylindrical Algebraic Decomposition (CAD) is a key tool in computational algebraic geometry, particularly for quantifier elimination over real-closed fields. However, it can be expensive, with worst case complexity do...
详细信息
ISBN:
(纸本)9781509057085
Cylindrical Algebraic Decomposition (CAD) is a key tool in computational algebraic geometry, particularly for quantifier elimination over real-closed fields. However, it can be expensive, with worst case complexity doubly exponential in the size of the input. Hence it is important to formulate the problem in the best manner for the CAD algorithm. One possibility is to precondition the input polynomials using Groebner Basis (GB) theory. Previous experiments have shown that while this can often be very beneficial to the CAD algorithm, for some problems it can significantly worsen the CAD performance. In the present paper we investigate whether machine learning, specifically a support vector machine (SVM), may be used to identify those CAD problems which benefit from GB preconditioning. We run experiments with over 1000 problems (many times larger than previous studies) and find that the machine learned choice does better than the human-made heuristic.
Textual entailment recognition (RTE) is the task of deciding, when given two text fragments, whether the meaning of one text is entailed (can be inferred) from the other text. Last year, we built our first Textual Ent...
详细信息
Textual entailment recognition (RTE) is the task of deciding, when given two text fragments, whether the meaning of one text is entailed (can be inferred) from the other text. Last year, we built our first Textual Entailment (TE) system, with which we participated in the RTE3 competition. the main idea of this system is to transform the hypothesis making use of extensive semantic knowledge from sources like DIRT, WordNet, Wikipedia and acronyms database. Additionally, the system applies complex grammar rules for rephrasing in English and uses the results of a module we built to acquire the extra background knowledge needed. In the first part, we presented the system architecture and the results, whose best run ranked 3 rd in RTE3 among 45 participating runs of 26 *** second part of the paper presents the manner in which we adapted the TE system in order to include it in a Question Answering (QA) system. the aim of using the TE system as a module in the general architecture of a QA system is to improve the ranking between possible answers for questions in which the answer type is Measure, Person, Location, Date and Organization.
暂无评论