Practical pattern-classification and knowledge-discovery problems require the selection of a subset attributes or features to represent patterns be classified. The authors' approach uses genetic algorithm select such subsets, achieving multicriteria optimization in terms generalization accuracy costs associated with features.

While some studies have proven that Swin Transformer (Swin) with window self-attention (WSA) is suitable for single image super-resolution (SR), the plain WSA ignores broad regions when reconstructing high-resolution images due to a limited receptive field. In addition, many deep learning SR methods suffer from intensive computations. To address these problems, we introduce N-Gram context low-level vision Transformers first time. We define as neighboring local windows in Swin, which differs...

Constructive learning algorithms offer an attractive approach for the incremental construction of near-minimal neural-network architectures pattern classification. They help overcome need ad hoc and often inappropriate choices network topology in that search suitable weights a priori fixed architectures. Several such are proposed literature shown to converge zero classification errors (under certain assumptions) on tasks involve binary mapping (i.e., problems involving binary-valued input...

Indoor air quality analysis is of interest to understand the abnormal atmospheric phenomena and external factors that affect quality. By recording analyzing measurements, we are able observe patterns in measurements predict near future. We designed a microchip made out sensors capable periodically proposed model estimates changes using deep learning. In addition, developed an efficient algorithm determine optimal observation period for accurate prediction. Experimental results with...

With the proliferation of Internet and huge amount data it transfers, text summarization is becoming more important. We present an approach to design automatic summarizer that generates a summary by extracting sentence segments. First, sentences are broken into segments special cue markers. Each segment represented set predefined features (e.g. location segment, average term frequencies words occurring in number title like). Then supervised learning algorithm used train extract important...

Abstract Motivation: Automatic knowledge discovery and efficient information access such as named entity recognition relation extraction between entities have recently become critical issues in the biomedical literature. However, inherent difficulty of task, mainly caused by diversity natural language, is further compounded domain because sentences are commonly long complex. In addition, often involves modeling range dependencies, discontiguous word patterns semantic relations for which...

The construction of interaction networks between proteins is central to understanding the underlying biological processes. However, since many useful relations are excluded in databases and remain hidden raw text, a study on automatic extraction from text important bioinformatics field. Here, we suggest two kinds kernel methods for genic extraction, considering structural aspects sentences. First, improve our prior dependency by modifying function so that it can involve various substructures...

Diffusion Transformers (DiT) excel in video generation but encounter significant computational challenges due to the quadratic complexity of attention. Notably, attention differences between adjacent diffusion steps follow a U-shaped pattern. Current methods leverage this property by caching blocks, however, they still struggle with sudden error spikes and large discrepancies. To address these issues, we propose UniCP unified pruning framework for efficient generation. optimizes both...

The automatic document metadata extraction process is animportant task in a world where thousands of documents are just one``click'' away. Thus, powerful indices necessary to support effective retrieval. upcoming XML standard represents an important step this direction as itssemistructuredrepresentation conveys together with the text document. For example, retrieval scientific papers by authors or affiliations would be straightforward tasks if were stored XML.Unfortunately, today, largest...

Multi-layer networks of threshold logic units (TLU) offer an attractive framework for the design pattern classification systems. A new constructive neural network learning algorithm (DistAl) based on inter-pattern distance is introduced. DistAl constructs a single hidden layer hyperspherical neurons. Each neuron designed to determine cluster training patterns belonging same class. The weights and thresholds neurons are determined directly by comparing distances patterns. This offers...

Although sex differences have been reported in patients with clear cell renal carcinoma (ccRCC), biological has not received clinical attention and genetic between sexes are poorly understood. This study aims to identify sex-specific gene mutations explore their significance ccRCC. We used data from The Cancer Genome Atlas-Kidney Renal Clear Cell Carcinoma (TCGA-KIRC), Cancer-European Union (RECA-EU) Korean-KIRC. A total of 68 sex-related genes were selected TCGA-KIRC through machine...

For their analgesic and anti-arthritic effects, Aconitum species have been used in folk medicine some East Asian countries. Although effect is attributed to its action on voltage-dependent sodium channels, they also suppress purinergic receptor expression dorsal root ganglion neurons rats with neuropathic pain. In vitro study demonstrated that the suppresses ATP-induced P2X7 (P2X7R)-mediated inflammatory responses microglial cell lines. Herein, we examined of intrathecal administration...

Tools for selective proactive as well reactive information retrieval and knowledge discovery constitute some of the key enabling technologies managing data overload translating recent advances in automated acquisition, digital storage, computers communications into fundamental decision support, scientific related applications. The paper describes an implementation intelligent, customizable mobile software agents from distributed sources. These tools are part network (DKN) toolbox that is...

In the field of machine learning and pattern recognition, feature subset selection is an important area, where many approaches have been proposed. this paper, we choose some algorithms analyze their performance using various datasets from public domain. We measured number reduced features improvement with chosen methods, then evaluated compared each method on basis these measurements.

Many feature subset selection algorithms have been proposed and discussed for years. However, the problem of finding optimal from full data still re- mains to be a difficult problem. In this paper, we propose novel methods find rele- vant by using biologically-inspired such as Genetic Algorithm Particle Swarm Optimization. We also variant approach considering significance each feature. verified performance experiments with various real-world datasets. Our based on produced better than other...

There is a strong positive correlation between the development of deep learning and amount public data available. Not all can be released in their raw form because risk to privacy related individuals. The main objective privacy-preserving publication anonymize while maintaining utility. In this paper, we propose semi-generative adversarial network (PPSGAN) that selectively adds noise class-independent features each image enable processed maintain its original class label. Our experiments on...

Cerebrovascular accidents (CVA) cause a range of impairments in coordination, such as spectrum walking ranging from mild gait imbalance to complete loss mobility. Patients with CVA need personalized approaches tailored their degree impairment for effective rehabilitation. This paper aims evaluate the validity using various machine learning (ML) and deep (DL) classification models (support vector machine, Decision Tree, Perceptron, Light Gradient Boosting Machine, AutoGluon, SuperTML, TabNet)...

In the field of machine learning and pattern recognition, feature subset selection is an important area, where many approaches have been proposed. this paper, we choose some algorithms analyze their performance using various datasets from public domain. We measured number reduced features improvement with chosen methods, then evaluated compared each method on basis these measurements.