Wind power output prediction is crucial to the balanced dispatch and market operation of system, accurate wind can guarantee security stability grid operation. In order realize highly support low-carbon transformation a method integrating convolutional neural network (CNN), bi-directional long short-term memory (BiLSTM) Attention mechanism proposed. To address nonlinear temporal characteristics data, CNN layer extracts local features, BiLSTM captures bidirectional dependencies, reduces...

A newly developed indicator that can be used to determine the variation in demand a power system withstand is presented. Unlike most indicators are currently available, new as forecasting tool. The results presented show it feasible utilize Electricite de France's national control center.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

For video summarization and retrieval, one of the important modules is to group temporal-spatial coherent shots into high-level semantic clips namely scene segmentation. In this paper, we propose a novel segmentation categorization approach using normalized graph cuts(NCuts). Starting from set shots, first calculate shot similarity key frames. Then by modeling as partition problem where each node weight edge represents between two employ NCuts find optimal automatically decide optimum number...

Complex shapes can be effectively analyzed by multiresolution shape descriptors. Compared with wavelet descriptors that are widely used for analysis, Fourier have better invariance properties and higher computational efficiency. We propose a novel scheme to generate analysis: downsampling expansion followed upsampling reconstruction. Simulation shows our outperforms both traditional in terms of accuracy

Currently, deep learning, especially convolutional neural network (CNN), has been widely applied in imaging and computer vision application due to the good accuracy image classification pattern recognition tasks. In this paper, a simple CNN model is built for heart diseases classification. This directly trained by images without using transfer learning of pre-trained networks. The networks such as AlexNet GoogLeNet have used processing applications. However, these are non-medical images;...

Zernike moments are widely used in shape retrieval, recognition and classification. The rotational invariance property of is very simple to achieve, due their separable magnitude-phase property. However, not directly invariant scale translation. Recently Cartesian invariants (CZMI) were introduced make under translation scale. Although CZMI reduced error considerably, they inconsistent increases for high aspect ratio images. In this paper, we propose a new parameter, which reduces errors,...

This paper introduces a domain specific knowledge discovery technique that is applicable for both information retrieval and text mining, identifying word meanings characterized by domains. The meaning of words identified using fusion algorithm not only narrows concepts from different but also avoids the unknown problem so domains can be found in series words. Domain presented purpose experiments on medical documents. Experiments performed over two fields: query expansion classification...

Archives of human rights violations reports, by virtue their poor metadata, basis in natural language, and scale, obscure fine grain analyses violation event patterns. Cross-document coreference victim or perpetrator occurrences from across a corpus is challenging, particularly when those mentions relate to different events. These challenges are emblematic the transition small scale big data analysis humanities. This paper discusses these issues proposes framework address so as explore...

The advent of the Internet and improvements in data sharing storage, have resulted an explosion textual data. But, complete assimilation such massive amounts its raw form is a daunting task. Automated text mining methods as summarization present user with condensed version containing only key information. This especially useful case online forums that contain large number posts spread out across several threads. Document been extensively studied developed recent past. paper aims at...

Advances in whole slide imaging technology have promoted a high use of digital images and generated large volume image data that is reliable useful determining treatment outcome. Recent technologies closely related to machine learning deep algorithms contributed the success histopathology by analyzing digitized providing quantitative information are for faster turnaround times effective patient. The histopathological analysis has received much attention due its capability mitigating problem...

The fluorescence spectrum of pesticides whose structures are very similar overlap in a certain wavelength range. To the classification and recognition overlapping spectrum, BP network has shortcomings slow training speed high error rate. An improved wavelet neural (WNN) is presented this paper. topology given, basis selected its algorithm designed to carry out design experimental system. By using WNN separately, simulation research carbofuran carbaryl been done. results show that higher...

For an application in volunteer computing environments, providing a reliable scheduling based on resource reliability evaluation is becoming increasingly important. Most existing reputation models used for ignore the task runtime influence. Moreover, to optimize makespan and workflow applications, most works use list heuristics rather than genetic algorithms (GAs) which can usually give better solutions. Hence, this paper, we propose look-ahead algorithm (LAGA) both time application. LAGA...

Biological data classification is an important mining research area in biomedical applications. The current challenge problem that there a large number of condition attributes (features) biological data, with which it difficult for methods to deal. In this paper, new approach based on rough sets and support vector machines proposed classification. Rough theory good mathematical tool make attribute reduction by removing redundant (features). Furthermore, the use information entropy as...

Time series analysis is vital for numerous applications, and transformers have become increasingly prominent in this domain. Leading methods customize the transformer architecture from NLP CV, utilizing a patching technique to convert continuous signals into segments. Yet, time data are uniquely challenging due significant distribution shifts intrinsic noise levels. To address these two challenges,we introduce Sparse Vector Quantized FFN-Free Transformer (Sparse-VQ). Our methodology...

The implementation of the two-carbon target "carbon peak" and neutrality". increase comprehensive energy micro-grid will form a cluster, namely multi-micro-grid integrated system. Compared with single system can greatly improve cost grid. How to further overall use efficiency is still key point research. In this paper, an optimal scheduling model established for multi-micro grid including electric interaction considering efficient utilization cascade. Firstly, architecture quality...

Time series forecasting has played a significant role in many practical fields. But time data generated from real-world applications always exhibits high variance and lots of noise, which makes it difficult to capture the inherent periodic patterns data, hurting prediction accuracy significantly. To address this issue, we propose Esiformer, apply interpolation on original decreasing overall alleviating influence noise. What's more, enhanced vanilla transformer with robust Sparse FFN. It can...

Deep learning algorithms have been successfully adopted to extract meaningful information from digital images, yet many of them untapped in the semantic image segmentation histopathology images. In this paper, we propose a deep convolutional neural network model that strengthens Atrous separable convolutions with high rate within spatial pyramid pooling for segmentation. A well-known called DeepLabV3Plus was used encoder and decoder process. ResNet50 block which provides us advantage...