Hasil untuk "Computer software"

F. Rohlf

Qi-yi Tang, Chuanquan Zhang

G. Stoet

L. Bass, P. Clements, R. Kazman

P. Wessel, Walter H. F. Smith

F. Mohamadi, N. Richards, W. Guida et al.

J. Thioulouse, D. Chessel, S. Dolédec et al.

W. Dupont, W. Plummer

Steven K. Johnson

U. Lorenzo-Seva, P. J. Ferrando

J. Jacko, Andrew Sears

R. Buyya, Shin Chee, Yeo et al.

J. Goldberg, Xerxes P. Kotval

B. Contreras-Moreira, P. Vinuesa

L. Danyushevsky, P. Plechov

Xin Ye, Hui Shen, Xiao Ma et al.

Per Larsen, Andrei Homescu, Stefan Brunthaler et al.

Hijab Zehra Zaidi, Ubaid Ullah, Muddassira Arshad et al.

This paper analyses existing research about machine learning approaches in software defect prediction as a key element for improving software reliability and quality. The paper reviews the use of machine learning algorithms in software defect prediction framework’s bug prediction while assessing their performance across multiple environments. A comprehensive review of scholarly literature enables researchers to specify both advantages and drawbacks that emerge when using machine learning for automated defect detection in software defect prediction applications. The review conducts assessments of typical metrics like accuracy and precision and recall and runtime performance yet extends its evaluation to analyze new trends combining deep learning with ensemble approaches to enhance software defect prediction functionality. The examined findings provide crucial guidelines which help developers select and improve machine learning models in software defect prediction processes that result in better software reliability and robustness.

Ganzorig Batnasan, Munkh-Erdene Otgonbold, Qurban Ali Memon et al.

Sign Language Recognition (SLR) presents a significant challenge as a fine-grained, scene- and subject-invariant video classification task, primarily relying on hand gestures and facial expressions to convey meaning. Vision foundation models, such as Vision Transformers (ViTs), trained on general human action recognition datasets, often struggle to capture the nuanced features of signs. We highlight two main challenges: 1) the loss of critical spatial features in the head and hand regions due to video downscaling during preprocessing, and 2) the lack of sufficient domain-specific knowledge of sign gestures in ViTs. To tackle these, we propose a pipeline comprising our Head & Hands Tunneling (H&HT) preprocessor and a domain-specifically pre-trained 32-frame ViT classifier. The H&HT preprocessor, incorporating the MediaPipe pose predictor, maximizes the preservation of critical spatial details from the signer’s head and hands in raw sign language videos. When the ViT model is pre-trained on a domain-specific, large-scale SLR dataset, the two parts complement each other. As a result, the 32-frame H&HT pipeline achieves a Top-1 accuracy of 62.82% on the WLASL2000 benchmark, surpassing the performance of the 32-frame models and ranking second among the 64-frame models. We also provide benchmarking results on the ASL-Citizen dataset and two revised versions of the WLASL2000 dataset. All weights and codes are available in this link.

Malte Pütz, Romain Vasseur, Andreas W.W. Ludwig et al.

In circuit-based quantum state preparation, qubit loss and coherent errors are circuit imperfections that imperil the formation of long-range entanglement beyond a certain threshold. The critical theory at the threshold is a continuous entanglement transition known to be described by a (2+0)-dimensional nonunitary conformal field theory which, for the two types of imperfections of certain circuits, is described by either percolation or Nishimori criticality, respectively. Here we study the threshold behavior when the two types of errors simultaneously occur and show that, when moving away from the Clifford regime of projective stabilizer measurements, the percolation critical point becomes unstable and the critical theory flows to Nishimori universality. We track this critical renormalization group (RG) crossover flow by mapping out the entanglement phase diagrams, parametrized by the probability and strength of random weak measurements, of two dual protocols preparing surface code or GHZ-class cat states from a parent cluster state via constant-depth circuits. Extensive numerical simulations, using hybrid Gaussian fermion and tensor network/Monte Carlo sampling techniques on systems with more than a million qubits, demonstrate that an infinitesimal deviation from the Clifford regime leads to a sudden, strongly nonmonotonic entanglement growth at the incipient nonunitary RG flow. We argue that spectra of scaling dimensions of both the percolation and Nishimori fixed points exhibit multifractality. For percolation, we provide the exact (nonquadratic) multifractal spectrum of exponents, while for the Nishimori fixed point we show high-precision numerical results for five leading exponents characterizing multifractality.