Early fault detection technique is crucial to reduce the machine downtime and has high impact on a wide variety of industrial applications. However, early still subject following challenges: 1) extracting features from incipient signals; 2) detecting anomalies with considering sequential data correlation; 3) enhancing reliability alarm. In this paper, we introduce novel deep-structured framework solve problem. First, system variation measured deviation value generated by current feature...

Training an artificial neural network with backpropagation algorithms to perform advanced machine learning tasks requires extensive computational process. This paper proposes implement the algorithm optically for in situ training of both linear and nonlinear diffractive optical networks, which enables acceleration speed improvement energy efficiency on core computing modules. We demonstrate that gradient a loss function respect weights layers can be accurately calculated by measuring forward...

Multiple image hiding aims to hide multiple secret images into a single cover image, and then recover all perfectly. Such high-capacity may easily lead contour shadows or color distortion, which makes very challenging task. In this paper, we propose novel framework based on invertible neural network, namely DeepMIH. Specifically, develop an network (IHNN) innovatively model the concealing revealing as its forward backward processes, making them fully coupled reversible. The IHNN is highly...

Quantitative identification of heavy metals (HM) sources in soils is key to prevention and control metal pollution. In this study, UNMIX, PMF (Positive matrix factorization) model Pb-Zn-Cu isotopic compositions were combined quantitatively identify a suburban agricultural area Kaifeng, China. Using multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS) ICP-MS, we measured Pb, Zn Cu stable compositions, HM concentrations chemical fractions studied soils, as well potential...

Separating singing voice from music accompaniment is very useful in many applications, such as lyrics recognition and alignment, singer identification, information retrieval. Although speech separation has been extensively studied for decades, little investigated. We propose a system to separate monaural recordings. Our consists of three stages. The detection stage partitions classifies an input into vocal nonvocal portions. For portions, the predominant pitch detects then uses detected...

This paper presents the design and construction of facade renovation project ("Bamboo Cubic" project) Huangqiao Square in Shaowu City, Fujian Province, China. In this project, structural form cross-sectional dimensions were determined using a combination manual finite element analysis to meet relevant regulations. Once was confirmed, primary components such as foundation, column base, connection between frame elements designed comply with requirements. Innovative connections used install...

Unmanned Aerial Vehicle (UAV) technology has been widely applied in both military and civilian applications. Recent researches on UAV systems feature the dramatic augment of variety number equipped sensors, which results such an issue that multiple UAVs cannot afford to handle big data generated by a range sensors air. Considering this practical problem, paper, we propose cloud-based system incorporates computing capability terrestrial cloud into systems. Relying proposed system, one...

Monaural musical sound separation has been extensively studied recently. An important problem in of pitched sounds is the estimation time-frequency regions where harmonics overlap. In this paper, we propose a sinusoidal modeling-based system that can effectively resolve overlapping harmonics. Our strategy based on observations same source have correlated amplitude envelopes and change phase harmonic related to instrument's pitch. We use these two least squares framework for The directly...

Traditional frame-based video frame interpolation (VFI) methods rely on the linear motion assumption and brightness invariance assumption, which may lead to fatal errors confronting scenarios with high-speed motions. To tackle above challenge, inspired by advantages of event cameras asynchronously recording changes at each pixel, we propose a Fast-Slow joint synthesis framework for event-enhanced interpolation, named SuperFast, in this paper, can generate high rate (5000 FPS, 200× faster)...

Quercetin (QR) is a naturally occurring flavonoid organic compound that has poor solubility in water and highly unstable alkaline conditions, resulting limited absorption poultry. Consequently, our experiment, QR was employed as model compound, encapsulated within the caffeic acid graft chitosan copolymer (CA-g-CS) self-assembled micelles to enhance its solubility, stability exhibit synergistic antibacterial effect. The optimization of formula carried out using combination single-factor...

Recent years have witnessed remarkable achievements in video-based action recognition. Apart from traditional frame-based cameras, event cameras are bio-inspired vision sensors that only record pixel-wise brightness changes rather than the value. However, little effort has been made event-based recognition, and large-scale public datasets also nearly unavailable. In this paper, we propose an recognition framework called EV-ACT. The Learnable Multi-Fused Representation (LMFR) is first...

Action recognition from video data forms a cornerstone with wide-ranging applications. Single-view action faces limitations due to its reliance on single viewpoint. In contrast, multi-view approaches capture complementary information various viewpoints for improved accuracy. Recently, event cameras have emerged as innovative bio-inspired sensors, leading advancements in event-based recognition. However, existing works predominantly focus single-view scenarios, leaving gap exploitation,...

Codebook-based learning provides a flexible way to extract the contents of an image in data-driven manner for visual recognition. One central task such frameworks is codeword assignment, which allocates local descriptors most similar codewords dictionary generate histogram categorization. Nevertheless, existing assignment approaches, e.g., nearest neighbors strategy (hard assignment) and Gaussian similarity (soft assignment), suffer from two problems: 1) too strong Euclidean assumption 2)...

Light helds suffer from a fundamental resolution tradeoff between the angular and spatial domain. In this paper, we present novel cross-scale light held super-resolution approach (up to 8× gap) super-resolve low-resolution (LR) images that are arranged around high-resolution (HR) reference image. To bridge enormous gap inputs, introduce an intermediate view denoted as single image (SISR) image, i.e., super-resolving LR input via based scheme, which owns identical HR yet lacks high-frequency...

Chromatographic-based approaches are most widely used for the analysis of enantiomers. Unambiguous results achieved through visualization each enantiomer, while separation process is often time consuming. By endowing 19F probes with suitable dynamic recognition properties, we develop here a detection scheme that produces chromatogram-like output in enantiomeric mixtures without separation. A wide range chiral compounds, including alcohols, ethers, amides, carbamates, oxazolidinones,...

It is widely acknowledged that biological intelligence capable of learning continually without forgetting previously learned skills. Unfortunately, it has been observed many artificial techniques, especially (deep) neural network (NN)-based ones, suffer from catastrophic problem, which severely forgets previous tasks when a new one. How to train NNs forgetting, termed continual learning, emerging as frontier topic and attracting considerable research interest. Inspired by memory replay...

The objective of this study was to create and authenticate a prognostic model for lymph node metastasis (LNM) in colorectal cancer (CRC) that integrates clinical, radiomics, deep transfer learning features.