Now at VMware. Multithreaded deterministic replay has important applications in cyclic debugging, fault tolerance and intrusion analysis. Memory race recording is a key technology for multithreaded replay. In this paper, we considerably improve our previous always-on Flight Data Recorder (FDR) four ways: •Longer by reducing the log size growth rate to approximately one byte per thousand dynamic instructions. •Lower hardware cost 24 KB processor core. •Simpler design modifying only cache...

While existing prediction models built on popular deep architectures have shown promising results in facial depression recognition, they still lack sufficient discriminative power due to the issues of 1) limited amount labeled data for representation learning and, 2) large variation expression across different persons same score and subtle difference levels. In this article, we formulate recognition as a label distribution (LDL) problem, propose joint metric (DJ-LDML) method address these...

Given a pair of facial images, it is an interesting yet challenging problem to determine if there kin relation between them. Recent research on that topic has made encouraging progress by learning similarity metric from kinship data. However, most the existing algorithms cannot handle hard samples very well, i.e., some ambiguous test pairs be well classified due compounding factors, such as large age gap or gender difference parents and children. To address this, we propose Adversarial...

Kinship verification via facial images is an emerging problem in computer vision and biometrics. Recent research has shown that learning a kin similarity measurement plays critical role constructing vision-based kinship system. We propose this paper new metric method for on human faces. To end, we first extract multiple feature representations each face image using different descriptors. Then, sparse bilinear models (one view) are jointly learned by joint structured sparsity-inducing norms,...

Motivated by the key observation that children generally resemble their parents more than other persons with respect to facial appearance, distance metric (similarity) learning has been dominant choice for state-of-the-art kinship verification via images in wild. Most existing learning-based approaches verification, however, are focused on a genetic similarity measure batch manner, leading less scalability practical applications ever-growing amount of data. To address this, we propose new...

In this paper, we investigate the problem of prediction confidence in face and kinship verification. Most existing verification methods focus on accuracy performance while ignoring estimation for their results. However, is essential modeling reliability trustworthiness such high-risk tasks. To address this, introduce an effective measure that allows models to convert a similarity score into any given pair. We further propose confidence-calibrated approach, termed Angular Scaling Calibration...

Existing optimization methods to heterogeneous redundancy allocation problem often suffer from the local-trap in optimization, due rugged energy landscapes. In this paper, a new paradigm based on Markov chain Monte Carlo sampling is proposed for solving multi-state systems. We address an optimization-by-sampling framework, and propose sample intricate distribution over combinatorial space by doubly adaptive approach, where target adaptation favors free random walk landscape substantially...

Vision-based kinship recognition aims to determine whether the face images have a kin relation. Compared traditional solutions, vision-based methods advantages of lower cost and being easy implement. Therefore, such technique can be widely employed in lots scenarios including missing children search automatic management family album. The Recognizing Families Wild (RFIW) Data Challenge provides platform for evaluation different approaches with ranked results. We propose supervised contrastive...

Illegal construction should be detected as early possible it can damage the environment and economy. However, existing methods for detecting illegal improved in terms of their detection cycles, accuracy, speed. Moreover, there are relatively few valuable real-world image datasets construction. To address these issues, a high-precision real-time model named YEMNet new large-scale dataset objects (ICOS) proposed herein. Our is based on You Only Look Once v3 object model; this adopts...

Based on a comprehensive analysis of different work and management indicators, as well the quantitative assessment outcomes implemented personal safety responsibility system, this study proposes utilization computer intelligent warning technology within power supply enterprise system. This enables automatic entire operational process facilitating examination system at various hierarchical levels organization. paper aims to propose establishment evaluation indicators derived from components...

With the rapid development of distribution networks and increasing demand for electricity, pressure power supply medium- low-voltage (M&LVDNs) is increasingly significant, especially considering large scale customers at (LV) level. In this paper, an outage sequence optimization method (LVDNs) that considers importance users proposed. The aims to develop optimal strategy LV in case medium-voltage (MV) failure events. First, a multi-dimensional indicator system constructed, are ranked...

This article proposes a panoramic mapping and interaction method for real-world images power grid twins. The data of the scene is obtained through high-precision acquisition technology, accurate positioning carried out. Secondly, collected point cloud optimized processed to construct twin model, texture performed. algorithm used map model image. Real time object detection tracking, attitude estimation, visual feedback effect presentation are YOLO Kalman filter technology achieve target...

The advent of Contrastive Language-Image Pre-training (CLIP) models has revolutionized the integration textual and visual representations, significantly enhancing interpretation static images. However, their application to video recognition poses unique challenges due inherent dynamism multimodal nature content, which includes temporal changes spatial details beyond capabilities traditional CLIP models. These necessitate an advanced approach capable comprehending complex interplay between...

With the needs of decision-support information enterprise and fast development computer technologies data warehouse technology come out. The is a repository collected from multiple, possibly heterogeneous, autonomous, distributed databases. stored at in form views referred to as materialized views. design one core research problems studying evolution warehouse. One most important decisions selection. Selecting materialize impacts on efficiency well total cost establishing running So, we...

In this paper, we propose a fast and accurate block-matching algorithm for motion estimation of human faces via Artificial Bee Colony (ABC) optimization. The mean square error (MSE) is often used as the matching metric in block matching, which, however, has high computational cost practice. By using ABC optimization, introduce novel effective metric. We develop based on proposed to improve accuracy with lower cost. Experimental results show that our method could achieve significant...

While encouraging results have been made so far to advance kinship verification by using facial images, learning a robust genetic similarity measure remains challenging, especially in the setting of general verification, wherein gender labels test samples are unknown advance. In this paper we present deep metric method with carefully designed two-stream neural network jointly learn pair embeddings for parent-child images. particular, first modeled explicitly consist common and individual...

Most existing approaches to heterogeneous redundancy allocation problem (RAP) are prone getting trapped in local optimal modes during optimization, mainly due the rugged combinatoric landscapes. Recently, optimization-by-sampling paradigm based on stochastic approximation Monte Carlo (SAMC) sampling has shown superior performance solving RAP for multistate systems (MSSs). However, one drawback of this method is that global move a Markov chain relying only uniform distribution typically hard...