This work deals with the system identification of Thevenin models of Li-Po batteries for UAV applications. Starting from the results of an experimental hybrid pulse-power characterization of a battery pack carried out...
This work deals with the system identification of Thevenin models of Li-Po batteries for UAV applications. Starting from the results of an experimental hybrid pulse-power characterization of a battery pack carried out at different temperatures (0°C, 15°C, 49°C) and within the operative range of state-of-charge (>10%), the model parameters are identified via three heuristic optimizationalgorithms, based on particle-swarm, teaching-learning and differential evolution techniques. Differently from conventional approaches typically applied by commercial CAE tools (e.g. Matlab), the proposed techniques are directly applied to the whole time history of the measurements. The results highlight that the particle-swarm method exhibits the fastest convergence, but it requires to initially define the algorithm weighing coefficients. This is not needed for teaching-learning based optimization, but computational effort strongly increases to achieve satisfactory accuracy. The differential evolution technique provides intermediate performances, especially if the total computation time is also considered. The case study is referred to the 1850 mAh/6 cells/22.2 V Li-Po battery pack employed in the lightweight fixed-wing UAV Rapier X-25, developed by Sky Eye Systems (Italy).
The field of graph data mining, one of the most important AI research areas, has been revolutionized by graph neural networks (GNNs), which benefit from training on real-world graph data with millions to billions of n...
详细信息
ISBN:
(纸本)9781450392365
The field of graph data mining, one of the most important AI research areas, has been revolutionized by graph neural networks (GNNs), which benefit from training on real-world graph data with millions to billions of nodes and links. Unfortunately, the training data and process of GNNs involving graphs beyond millions of nodes are extremely costly on a centralized server, if not impossible. Moreover, due to the increasing concerns about data privacy, emerging data from realistic applications are naturally fragmented, forming distributed private graphs of multiple "data silos", among which direct transferring of data is forbidden. The nascent field of federated learning (FL), which aims to enable individual clients to jointly train their models while keeping their local data decentralized and completely private, is a promising paradigm for large-scale distributed and private training of GNNs. FedGraph2022 aims to bring together researchers from different backgrounds with a common interest in how to extend current FL algorithms to operate with graph data models such as GNNs. FL is an extremely hot topic of large commercial interest and has been intensively explored for machine learning with visual and textual data. The exploration from graph mining researchers and industrial practitioners is timely catching up just recently. There are many unexplored challenges and opportunities, which urges the establishment of an organized and open community to collaboratively advance the science behind it. The prospective participants of this workshop will include researchers and practitioners from both graph mining and federated learning communities, whose interests include, but are not limited to: graph analysis and mining, heterogeneous network modeling, complex data mining, large-scale machine learning, distributed systems, optimization, meta-learning, reinforcement learning, privacy, robustness, explainability, fairness, ethics, and trustworthiness.
Structural variants (SVs) are rearrangements of regions in an individual’s genome signal. SVs are an important source of genetic diversity and disease in humans and other mammalian species. The SV detection process i...
Structural variants (SVs) are rearrangements of regions in an individual’s genome signal. SVs are an important source of genetic diversity and disease in humans and other mammalian species. The SV detection process is susceptible to sequencing and mapping errors, especially when the average number of reads supporting each variant is low (i.e. low-coverage settings), which leads to high false-positive rates. Besides their rarity in the human genome, they are shared between related individuals. Thus, it’s advantageous to devise algorithms that focus on close relatives. In this paper, we develop a constrained-optimization method to detect germline SVs in genetic signals by considering multiple related people. First, we exploit familial relationships by considering a biologically realistic scenario of three generations of related individuals (a grandparent, a parent, and a child). Second, we pose the problem as a constrained optimization problem regularized by a sparsity-promoting penalty. Our framework demonstrates improvements in predicting SVs in related individuals and uncovering true SVs from false positives on both simulated and real genetic signals from the 1000 Genomes Project with low coverage. Further, our block-coordinate descent approach produces results with equal accuracy to the 3D projections of the solution, demonstrating feasibility for more complex and higher-dimensional pedigrees.
The goal of the Seventeenth internationalworkshop on Automatic Performance Tuning (iWAPT2022) is to bring together researchers who are investigating automated techniques for constructing and/or adapting algorithms an...
详细信息
ISBN:
(数字)9781665497473
ISBN:
(纸本)9781665497480
The goal of the Seventeenth internationalworkshop on Automatic Performance Tuning (iWAPT2022) is to bring together researchers who are investigating automated techniques for constructing and/or adapting algorithms and software for high-performance on modern complex machine architectures. iWAPT is a series of workshops that focus on research and techniques related to performance sustainability issues. The series provides an opportunity for researchers and users of automatic performance tuning (AT) technologies to exchange ideas and experiences acquired when applying such technologies to improve the performance of algorithms, libraries, and applications; in particular, on cutting edge computing platforms. The half-day workshops consist of presentations of research papers. Topics of interest include performance modeling; adaptive algorithms; autotuned numerical algorithms; libraries and scientific applications; empirical compilation; automated code generation; frameworks and theories of AT and software optimization; autonomic computing; and context-aware computing.
The article proposes information technology for solving the discrete optimization problems in Geographic Information Systems (GIS). The information model of geographic data (geodata) was further developed, which consi...
详细信息
ISBN:
(纸本)9781665426053
The article proposes information technology for solving the discrete optimization problems in Geographic Information Systems (GIS). The information model of geographic data (geodata) was further developed, which consists of a formalized combination of their spatial and attributive components. It takes into account relational, semantic and frame models for representing the knowledge of the attributive component in order to unify the branch geodata. The conceptual model of decision support in sectoral GIS has been further developed, which takes into account the information model of geodata and the dynamics of their change, the hierarchy of tasks of sectoral GIS, the input operational data, the function of preferences and the criterion of decision-making. A new method of decision support for GIS construction was developed on the base of the proposed conceptual model. The method of selection and determination of free parameters of swarm intelligence algorithms is developed to increase the effectiveness of GIS. It is suggested to determine the free parameters for individual swarm algorithms based on machine learning with reinforcement, namely the Q-Learning method. Based on this method Markov chains for the swarm algorithms were constructed. Reinforcement consisted in the expert analysis of the results obtained by a certain swarm algorithm. On the example of territorial administration, optimal values of the parameters for individual swarm algorithms were found.
Wav2vec2 self-supervised multilingual training learns speech units common to multiple languages, leading to better generalization capacity. However, Wav2vec2 is larger than other E2E ASR models such as the Conformer A...
Wav2vec2 self-supervised multilingual training learns speech units common to multiple languages, leading to better generalization capacity. However, Wav2vec2 is larger than other E2E ASR models such as the Conformer ASR. Therefore, the objective of this work is to reduce the Wav2vec footprint by pruning lines from the intermediate dense layers of the encoder block, since they represent about two thirds of the encoder parameters. We apply Genetic algorithms (GA) to solve the combinatorial optimization problem associated with pruning, which means running many copies of the Wav2vec2 decoder in parallel using multiprocessing on a computer grid, so an effort was made to optimize the GA for good performance with few CPUs. The experiments show a small absolute word error rate damage of 0.21% (1.26% relative) for a pruning of 40% and compare this value with those of the usual L1-norm pruning and model restructuring by singular value decomposition.
Recent advances and applications of machine learning algorithms are becoming more common in different fields. It is expected that some applications require the processing of large datasets with those algorithms, which...
详细信息
ISBN:
(纸本)9781450384414
Recent advances and applications of machine learning algorithms are becoming more common in different fields. It is expected that some applications require the processing of large datasets with those algorithms, which leads to high computational costs. Massively parallel GPU methods can be applied to surpass this limitation and reduce the execution time of these algorithms. The construction of approximate K-Nearest Neighbor Graphs (K-NNG) is frequently required for similarity search or other applications such as the t-SNE dimensionality reduction technique. The K-NNG represents the K closest points (neighbors) for each point in a set. In this paper, we propose and analyze an all-points K-Nearest Neighbor Graph construction algorithm on GPU called Warp-centric K-NNG (w-KNNG), which is based on the Random Projection Forest method. Usually, the construction or search for k-NN sets for high dimensional points presents challenges for its implementation on many-core processing units, due to the space limitation in maintaining these sets in high speed shared memory. We present three warp-centric approaches for our algorithm that efficiently search and maintain the k-NN high dimensional point sets in global memory. In our experiments, the new methods allows the algorithm to achieve up to 639% faster execution when compared to the state-of-the-art FAISS library, considering an equivalent accuracy of approximate K-NNG. One of the new strategies (w-KNNG atomic) is more successful when applied to a smaller number of dimensions, while the tiled w-KNNG approach was successful in general scenarios for higher dimensional points.
In the case of limited radar resources, the resource scheduling strategy is of great significance to improve the space target cataloging ability. The traditional radar resource scheduling strategy uses fixed rules to ...
详细信息
In the case of limited radar resources, the resource scheduling strategy is of great significance to improve the space target cataloging ability. The traditional radar resource scheduling strategy uses fixed rules to allocate radar resources for observation requirements, which cannot maximize radar resources in practical *** this paper, a radar resource scheduling optimization method based on improved ant colony algorithm is proposed, which quantifies the space target observation requirements weighted by the task priority, and optimizes the scheduling result with the highest requirement satisfaction, so that the radar resources can be optimally allocated and the ability of space observation be improved.
Hyperspectral image analysis is one of the most important topics in the field of remote sensing. As the band dimension of hyperspectral image increases, it may cause a curse of dimensionality. Band selection could cho...
详细信息
ISBN:
(纸本)9781665426053
Hyperspectral image analysis is one of the most important topics in the field of remote sensing. As the band dimension of hyperspectral image increases, it may cause a curse of dimensionality. Band selection could choose a subset of bands that is the most effective for classification recognition to achieve dimensionality reduction. Furthermore, binary coding is more suitable for solving the band selection problem. Therefore, a band selection method for hyperspectral image based on binary coded hybrid rice optimization algorithm is proposed in the paper. The band selection problem is solved as a combinatorial optimization problem by defining an objective function based on the classification accuracy and the number of bands. The proposed method is compared with other nature-inspired algorithms on Indian Pines, Salinas, Kennedy Space Center, and Pavia University hyperspectral datasets. Experimental results demonstrate that the proposed method achieves satisfactory results in terms of performance and execution time for band selection.
Electricity plays an indispensable role in human lives. Due the increasing need for electricity in domestic, commercial and industrial applications and the deletion of conventional sources,the power generation system ...
详细信息
ISBN:
(数字)9781728146850
ISBN:
(纸本)9781728146850
Electricity plays an indispensable role in human lives. Due the increasing need for electricity in domestic, commercial and industrial applications and the deletion of conventional sources,the power generation system is switched on to systems with renewable energy sources. Therefore the power quality problems arises and research has been going on to improve the power quality. This paper is a study about the various power quality improving algorithms applied to the hybrid wind solar power generation with multilevel inverters. in comparison with various optimizationalgorithms, more control parameters with a new algorithm called Rider optimization Algorithm (ROA) is suggested.
暂无评论