Real-world datasets often comprise outliers (e.g., due to operational error, intrinsic variability of the measurements, recording mistakes, etc.) and, hence, require cleansing as a prerequisite any meaningful machine learning analysis. However, data is laborious task that requires intuition or expert knowledge. In particular, selecting an outlier detection algorithm challenging this choice dataset-specific and depends on nature considered dataset. These difficulties have prevented...

Aiming to solve the problems of false detection, missed and insufficient detection ability infrared vehicle images, an target algorithm based on improved YOLOv5 is proposed. The article analyzes image characteristics then discusses in detail. uses DenseBlock module increase shallow feature extraction. Ghost convolution layer used replace ordinary layer, which increases redundant graph linear calculation, improves network extraction ability, amount information from original image. accuracy...

Abstract Despite previous efforts to map the proportioning of a concrete its strength, robust knowledge-based model enabling accurate strength predictions is still lacking. As an alternative physical or chemical-based models, data-driven machine learning methods offer promising pathway address this problem. Although can infer complex, non-linear, non-additive relationship between mixture proportions and large datasets are needed robustly train such models. This concern as reliable data...

Automatic modulation recognition (AMR) plays an essential role in modern communication systems. In recent years, various algorithms based on deep learning have been emerging, but the problem of low accuracy has not solved well. To solve this problem, existing MCLDNN algorithm, paper, we proposed improved spatiotemporal multi-channel network (IQ-related features Multi-channel Convolutional Bi-LSTM with Gaussian noise, IQGMCL). Firstly, dividing input IQ signals into three channels, time...

The human visual attention system plays an important role in infrared target recognition because it can quickly and accurately recognize small targets has good scene adaptability. This paper proposes detection method based on mechanism, which consists of three modules: a bottom-up passive module, top-down active decision feedback equalization. In the given Gaussian characteristics targets, idea combining knowledge-experience shape features is applied to implement feature extraction,...

Abstract An environmental procedure to extract titanium components and metallic iron from Ti-bearing blast furnace slag is accomplished via three steps, which are high-temperature modification, gravity separation hydrometallurgy method. The behaviors of during the modification process studied. feasibility separating rutile matrix phase investigated; based on analysis results, experiment carried out in order improve TiO

Previous studies have shown that the fatigue life distribution of metal materials fabricated with Additive Manufacturing (AM) methods, such as Direct Energy Deposited (DED) Ti-6.5Al-2Zr-1Mo-1V alloys, exhibits two peaks. To promote application AM in aerospace and other engineering fields, developing a strength evaluation method suitable for based on their unique behaviours distributions is necessary. In this paper, novel Detail Fatigue Rating (DFR) was developed to evaluate DED bimodal...

Automatic modulation recognition (AMR) is a critical technology in spatial cognitive radio (SCR), and building high-performance AMR model can achieve high classification accuracy of signals. problem essentially, deep learning has achieved excellent performance various tasks. In recent years, joint multiple networks become increasingly popular. complex wireless environments, there are signal types diversity characteristics between different Also, the existence interference environment makes...

Reconstruction techniques for communication signals represent a significant research focus within the field of signal processing. To overcome difficulty and low precision in reconstructing OFDM signals, we introduce reconstruction technique called TOR-GAN (Transformer-Based Signal GAN). Reconstructing IQ sequences using CNN RNN presents challenges capturing correlations between two signals. tackle this issue, VIT (vision transformer) approach was introduced into discriminator network. The is...

In order to select fault feature parameters simply and quickly improve the identification rate of diesel engine faults by using vibration signals, this paper proposes a method on basis Pearson correlation coefficient diagram (PCC Diagram) orthogonal signals. At first, acceleration signals are synchronously acquired in direction top side cylinder head. And time‐domain extracted from obtain (PCC). Then, used do parameter screening is constructed selecting with more than 0.9. Finally,...

As the representative of flexibility in optical imaging media, recent years, fiber bundles have emerged as a promising architecture development compact visual systems. Dedicated to tackling problems universal honeycomb artifacts and low signal-to-noise ratio (SNR) bundles, iterative super-resolution reconstruction network based on physical model is proposed. Under constraint solving two subproblems data fidelity prior regularization term alternately, can efficiently "regenerate" lost spatial...

most YOLO object detection neural networks prefer to focus on traditional RGB image, but previous studies rarely consider special network with compact architecture for infrared image. In this paper, we analyze original architecture, and propose a based by using different blocks from work small target We use layer, GhostConv convolution, Focus structure, Focal EIOU Loss soft NMS modules improve which improves the accuracy speed of The experimental results show that model can be effectively...

In GaN-based high-electron mobility transistors, although excellent electron confinement has been demonstrated using a graded AlGaN buffer with linearly decreasing Al-content along [0001] direction, guidelines for design are still lacking. To obtain overall pictures of the carrier distribution and energy-band profile in AlGaN/GaN/graded heterostructures, influences related physical parameters studied by one-dimensional self-consistent simulation. The results show that negative polarization...

in recent years, Deep Neural Network (DNN) based methods have achieved great success for changing inflexible machine to intelligent and live system. Internet of Things (IoT) applications equipped with DNN domains such as self-driving, speech processing video surveillance particular challenges. Since DNNs are memory-/computing- intensive applications, FPGAs customized accelerate on edge system due their low latency high energy efficiency. However, embedded CPU is usually essential IoT single...

The traditional object detection algorithm is difficult to extract its characteristic information due own features such as low resolution and small coverage area of objects, resulting in the inability achieve effective reliable recognition accuracy. Aiming at problem detection, this paper proposes a method visible light based on deep learning YOLOv5 algorithm. First all, total 4000 dataset created noon under background sunny cloudy weather, then used for training, which mAP@0.5 100 200 times...

Connected component (CC) labeling is time consuming during image segmentation and object identification, especially when input images are large-scale binary converting uses multi-threshold. So in this paper, we present a fast multi-level CC by combining online threshold one-pass algorithm, also detailed architecture for hardware implementation proposed. During labeling, firstly gray converted to several data flows with different thresholds. Then they marked run-length code respectively 2× 2...