The proceedings contain 141 papers. The topics discussed include: three- dimensional synthetic aperture radar image formation;R-LRCBNet: a YOLOV8-based oriented ship detection method of SAR image;multistatic, real-tim...
ISBN:
(纸本)9781733509671
The proceedings contain 141 papers. The topics discussed include: three- dimensional synthetic aperture radar image formation;R-LRCBNet: a YOLOV8-based oriented ship detection method of SAR image;multistatic, real-time, high resolution lunar penetrating imaging approach for water-ice detection;simulating advanced driver assistance systems using a high-fidelity physics-based asymptotic shooting and bouncing ray-tracing electromagnetics solver;on the suitability of the Julia programming language for computational electromagnetics;GPU acceleration using CUDA for computational electromagnetics;MESTI: Maxwell's equations solver with thousands of inputs;computing derivatives of finite element solutions by automatic derivative programs;a numerical study on EMI shielding strategies for LV battery pack components;metasurface characterization for low frequency shielding purposes;and simulations of the 5G beamforming performance for mm wave bands in an urban scenario.
Over the years, several systematic literature reviews have been published reporting advances in tools and techniques for automated assessment in Computer Science. However, there is not yet a major bibliometric study t...
详细信息
ISBN:
(纸本)9783031330223;9783031330230
Over the years, several systematic literature reviews have been published reporting advances in tools and techniques for automated assessment in Computer Science. However, there is not yet a major bibliometric study that examines the relationships and influence of publications, authors, and journals to make these research trends visible. This paper presents a bibliometric study of automated assessment of programming exercises, including a descriptive analysis using various bibliometric measures and data visualizations. The data was collected from the Web of Science Core Collection. The obtained results allow us to identify the most influential authors and their affiliations, monitor the evolution of publications and citations, establish relationships between emerging themes in publications, discover research trends, and more. This paper provides a deeper knowledge of the literature and facilitates future researchers to start in this field.
This paper proposes a framework for model learning and action planning of cloth manipulation tasks based on such high-dimensional and redundant sensor observations as raw pixel images. This framework is called the var...
详细信息
ISBN:
(纸本)9781665445405
This paper proposes a framework for model learning and action planning of cloth manipulation tasks based on such high-dimensional and redundant sensor observations as raw pixel images. This framework is called the variationally autoencoded dynamic policy programming (VAE-DPP), which is a combination of variationally autoencoded hidden Markov decision process (VAE-HMDP) and dynamic policy programming (DPP). At first, VAE-HMDP is learned, which consists of two disentangled latent spaces: for the image with VAE and for the hidden Markov decision process (HMDP). Then DPP is applied to the learned HMDP to plan action sequences efficiently. After investigating its effectiveness through a simple simulation of a toy problem, we applied VAE-DPP to two robotic cloth-folding tasks: (1) folding a handkerchief and (2) folding trousers. We successfully generated appropriate actions that reached the target folding state.
With the advent and the growing usage of Machine Learning as a Service (MLaaS), cloud and network systems are now offering the possibility to deploy ML tasks on heterogeneous clusters. Then, network and cloud operator...
详细信息
ISBN:
(纸本)9798350395679;9798350395662
With the advent and the growing usage of Machine Learning as a Service (MLaaS), cloud and network systems are now offering the possibility to deploy ML tasks on heterogeneous clusters. Then, network and cloud operators have to schedule these tasks, determining both when and on which devices to execute them. In parallel, several solutions, such as neural network compression, were proposed to build small models which can run on limited hardware. These solutions allow choosing the model size at inference time for any targeted processing time without having to re-train the network. In this work, we consider the Deadline Scheduling with Compressible Tasks (DSCT) problem: a novel scheduling problem with task deadlines where the tasks can be compressed. Each task can be executed with a certain compression, presenting a trade-off between its compression level (and, its processing time) and its obtained utility. The objective is to maximize the tasks utilities. We propose an approximation algorithm with proved guarantees to solve the problem. We validate its efficiency with extensive simulation, obtaining near optimal results. As application scenario, we study the problem when the tasks are Deep Learning classification jobs, and the objective is to maximize their global accuracy, but we believe that this new framework and solutions apply to a wide range of application cases.
The Extended Berkeley Packet Filter (eBPF) is a sandboxed virtual machine that runs on operating systems with kernel privileges. Currently, eBPF programs are either translated from a subset of C, called Restricted C, ...
详细信息
Memristor-based crossbars, which can achieve 1-2 orders of magnitude energy efficiency improvement over digital machines, have been introduced to accelerate the neural networks of machine learning tasks. Due to the hi...
详细信息
ISBN:
(纸本)9781665484855
Memristor-based crossbars, which can achieve 1-2 orders of magnitude energy efficiency improvement over digital machines, have been introduced to accelerate the neural networks of machine learning tasks. Due to the high voltage pulses repeatedly applied onto memristors during programming and online tuning, the effective resistance ranges of the memristors actually decrease as a result of aging, which eventually impair the inference accuracy of the neural network running on the memristor-based crossbar. In this paper, we propose an algorithmhardware co-design framework combining aging aware retraining and gradient sparsification to mitigate the impact of aging and extend the lifetime of the crossbar. Experimental results show that the proposed method can effectively increase the inference accuracy by up to 16% even with severe aging, while the crossbar lifetime can be extended by up to 2.7 x .
This paper focuses on the control of a fuel cell system, for which an optimization model is proposed to control the oxygen excess ratio in order to tackle the fuel cell starvation phenomenon and increase the net power...
详细信息
This paper focuses on the control of a fuel cell system, for which an optimization model is proposed to control the oxygen excess ratio in order to tackle the fuel cell starvation phenomenon and increase the net power production of the system. The optimization problem considers the detailed nonlinear dynamics of the stack and the air supply sub -system as constraints and upper and lower bounds to consider the technical limitations in the system's operation. The goal is to maintain the oxygen excess ratio at its reference value, using the compressor input voltage as a control variable, under several load variations of the current demand, considered as the external noise to the system. The proposed nonlinear optimization problem has been applied to a 75 kW stack used in the Ford P2000 fuel cell prototype vehicles, and results have been compared to the proportional -integral (PI) control technique. The proposed control approach has shown a significant enhancement in the control performance of the oxygen excess ratio.
Specifying and implementing a proof system from scratch requires significant effort. Logical Frameworks and Higher Order Logic programming Languages provide dedicated, high-level meta languages to facilitate this task...
详细信息
ISBN:
(纸本)9798400709692
Specifying and implementing a proof system from scratch requires significant effort. Logical Frameworks and Higher Order Logic programming Languages provide dedicated, high-level meta languages to facilitate this task in two ways: 1) variable binding and substitution are for free when meta language binders represent object logic ones;2) proof construction, and proof search, are greatly simplified by leveraging the unification procedure provided by the meta language. Notable examples of meta languages are Elf [21], Twelf [23], lambda Prolog [16], Beluga [24], Abella [8] and Isabelle [31] which have been used to implement or specify many formal systems such as First Order Logic [5], Set Theory [20], Higher Order Logic [19], and the Calculus of Constructions [4]. The object logic we are interested in is Coq's type theory [28]. We aim to develop a higher-order unification-based proof search procedure using the meta language Elpi [3], a dialect of lambda Prolog. Elpi's equational theory includes beta eta-equivalence and features a higher-order unification procedure similar or equal to(m) for the pattern fragment [15]. Elpi offers an encoding of Coq terms that is suitable for meta programming [6, 9, 26, 27] but that restricts similar or equal to(m) to first-order unification problems only. We refer to this basic encoding as O. In this paper we translate unification problems in O to an alternative encoding called M, from which we derive similar or equal to(O), the higher-order unification procedure of O. similar or equal to(O) honours beta eta-equivalence for terms within the pattern fragment, and allows for the use of heuristics when the terms fall outside the pattern fragment. Moreover, as similar or equal to(O) delegates most of the work to similar or equal to(m), it can be used to efficiently simulate a logic program in O by taking advantage of unification-related optimizations of the meta language, such as clause indexing.
We must retain liveness and exploratory practices within the programmingsystems that make us feel most productive. However, the temptation to just reuse black boxes through limited interfaces is pervasive. We expect ...
详细信息
ISBN:
(纸本)9781450399098
We must retain liveness and exploratory practices within the programmingsystems that make us feel most productive. However, the temptation to just reuse black boxes through limited interfaces is pervasive. We expect time savings and better performance at the cost of poor repairability. Fortunately, we also know about the benefits of having an open implementation constructed from familiar materials, integrated with familiar tools. Consequently, it is primarily a matter of "just building it" ... again? Piece of cake. What could possibly go wrong?
In this paper, we present a tool for monitoring the traces of cyber-physical systems (CPS) at runtime, with respect to Signal Temporal Logic (STL) specifications. Our tool is based on the recent advances of causation ...
详细信息
ISBN:
(纸本)9783031711763;9783031711770
In this paper, we present a tool for monitoring the traces of cyber-physical systems (CPS) at runtime, with respect to Signal Temporal Logic (STL) specifications. Our tool is based on the recent advances of causation monitoring, which reports not only whether an executing trace violates the specification, but also how relevant the increment of the trace at each instant is to the specification violation. In this way, it can deliver more information about system evolution than classic online robust monitors. Moreover, by adapting two dynamic programming strategies, our implementation significantly improves the efficiency of causation monitoring, allowing its deployment in practice. The tool is implemented as a C++ executable and can be easily adapted to monitor CPS in different formalisms. We evaluate the efficiency of the proposed monitoring tool, and demonstrate its superiority over existing robust monitors in terms of the information it can deliver about system evolution.
暂无评论