Vision Transformer (ViT) has shown promise in computer vision thanks to their powerful self-attention mechanisms. However, deploying ViT, especially for medical image analysis, presents significant challenges. Firstly, the quadratic computational complexity of self-attention hinders efficient model deployment. Secondly, medical images often have a lower signal-to-noise ratio (SNR) compared to natural images, making robust analysis difficult for standard ViTs. To address these specific limitations, we propose SparseMambaVision, a novel sparse-attention based Mamba-Transformer framework for prostate segmentation. Also, we introduce a novel Mamba Sparse Attention (MSA) module, which is designed to enhance both computational efficiency and model effectiveness to improve the performance on low SNR images. Extensive experiments on three benchmark prostate datasets demonstrate that SparseMambaVision achieves state-of-the-art results across four evaluation metrics, surpassing existing methods. This underscores the potential of SparseMambaVision for more accurate and efficient prostate segmentation in the clinical setting. The code is available at https://github.com/WangyuFeng-FJNU/SparseMambaVision.

Understanding how a trained deep neural network (DNN) classifies input data is critical for interpretability and trust in AI systems. Existing tools such as SplineCam can visualize theoretically exact decision boundaries for neural networks with piecewise polynomial activation functions. However, these methods often struggle with more commonly used smooth activations such as Sigmoid, SiLU, and their combinations. In this paper, we introduce a novel framework termed SigmaCam using a recursive algorithm that is both computationally efficient and capable of generating theoretically exact 2D decision boundaries for any Multi-Layer Perceptron (MLP) employing smooth activations. Our approach extends previous works to a broader class of activation functions and allows for the visualization of decision boundaries on the input domain of any dimension, analogous to the capabilities of SplineCam, but now applicable to widely used smooth nonlinearities. We demonstrate the effectiveness of our method by applying it to a variety of network architectures, activation function combinations (e.g., Sigmoid and SiLU), and datasets. The experiments show that our method can accurately capture complex decision boundaries, providing insights into model behavior and aiding in model interpretability. The proposed algorithm enables the analysis of networks that were previously difficult to handle, broadening the scope of exact decision boundary visualization tools beyond piecewise polynomial activations.

Alcohol Use Disorder (AUD), encompassing Alcohol Abuse (AA) and Alcohol Dependence (AD), is a chronic, relapsing brain condition that can cause mental and physical issues. Confusion between AA and AD can lead to ineffective or even excessively aggressive treatments, but distinguishing them is difficult due to similar symptoms and physiological indicators. Fortunately, from a psychological perspective, AA is closely linked to poor behavioral control, while AD is associated with affective lability. These psychological mechanisms are differently featured in the activity of patients' brain regions. Inspired by this, we propose a model called Multiscale and Hybrid Spatial Temporal Graph Convolutional Network (MuST-GCN) that enables to extract the features of electroencephalogram (EEG) signals for accurate identification of AA and AD within AUD. It includes two modules: Multiscale Feature Extraction (MSFE) module with GCN concurrently analyzes inter-regional and intra-regional connectivity of brain regions to extract EEG features, exploring functional connectivity differences in AA and AD patients; Hybrid Spatial Temporal Memory (HSTM) module integrates spatial and temporal attention mechanisms to refine the significant features extracted from MSFE, targeting the key brain regions and temporal dynamics. The HSTM module yields a refined feature representation reducing overfitting and improving multiclass classification accuracy. MuST-GCN is evaluated using five-fold cross-validation on two datasets, achieving classification accuracies of 90.74% and 99.99%, respectively, and demonstrating superior performance in identifying AA, AD, and AUD compared to existing methods.

Mental stress is an increasingly common psychological issue leading to diseases such as depression, addiction, and heart attack. In this study, an early detection framework based on electroencephalogram (EEG) data is developed for reducing the risk of these diseases. In existing frameworks, signals are often segmented into smaller sections prior to being input to a deep neural network. However, this approach ignores the fundamental nature of EEG signals as a carrier of valuable information (e.g., the integrity of frequency and phase, and temporal fluctuations of EEG components). As such, this type of segmenting may lead to information loss and a failure to effectively identify mental stress levels. Thus, we propose a novel multiclass classification framework termed multibranch LSTM and hierarchical temporal attention (MuLHiTA) for the early identification of mental stress levels. It specifically focuses on not only intraslice (within each slice) but also interslice (between different slices) samples in parallel. This was achieved by including two complementary branches, each of which integrated a specifically designed attention module into a bidirectional long short-term memory (BLSTM) network, enabling extraction of the most discriminative features from interslice and intraslice EEG signals simultaneously. The outputs of attention modules were then summed to obtain a feature representation that contributes to reduce overfitting and more effective multiclass classification. In addition, electrode positions were optimized using neural activity areas under high-stress conditions, thereby reducing computational costs by minimizing the number of critical electrodes. MuLHiTA was evaluated across one private [Montreal imaging stress task (MIST)] and two publicly available EEG datasets [EEG during mental arithmetic tasks (DMAT) and Simultaneous task EEG workload (STEW)]. These were divided into training and test sets using an 8:2 ratio, and the training data were further divided into training and validation sets using a fivefold cross-validation (CV) method, in which the model with the highest accuracy among the five was selected. The model was trained once more with the full training set, and the test data were then used to evaluate its performance. This approach achieved average classification accuracies of 93.58%, 91.80%, and 99.71% for the MIST, STEW, and DMAT datasets, respectively. Experimental results showed MuLHiTA was superior to state-of-the-art algorithms, including EEGNet, BLSTM, EEGLearn, convolutional neural network (CNN)-long short-term memory (LSTM), and convolutional recurrent attention model (CRAM), for multiclass classification. This demonstrates the viability of MuLHiTA for the early detection of mental stress.

Over the past decade, vision-based vehicle detection techniques for road safety improvement have gained an increasing amount of attention. Unfortunately, the techniques suffer from robustness due to huge variability in vehicle shape (particularly for motorcycles), cluttered environment, various illumination conditions, and driving behavior. In this paper, we provide a comprehensive survey in a systematic approach about the state-of-the-art on-road vision-based vehicle detection and tracking systems for collision avoidance systems (CASs). This paper is structured based on a vehicle detection processes starting from sensor selection to vehicle detection and tracking. Techniques in each process/step are reviewed and analyzed individually. Two main contributions in this paper are the following: survey on motorcycle detection techniques and the sensor comparison in terms of cost and range parameters. Finally, the survey provides an optimal choice with a low cost and reliable CAS design in vehicle industries.

Automated emergency braking (AEB) systems become more and more important than ever in modern vehicles for assisting drivers in emergency driving situations. They mostly require fusion techniques for vehicle detection (camera and radar or stereo-vision system) that require complicated algorithms and additional costs. These have caused AEB systems less attractive to the market. This paper presents an automobile detection algorithm using single camera for the AEB system. The algorithm contains three main steps: background subtraction, thresholding, and inverted U-shape back wheel detection. The simulation under MATLAB environment provides 87.25% and 78% of detection rate and accuracy, respectively for a 1080x1920 pixel input image; 88.25% and 73.5% of detection rate and accuracy for a 480x640 pixel input image. Processing time achieved are 0.156s and 0.0297s accordingly.

We introduce the concept of two dimensional (2D) scalability of trajectory piecewise linear (TPWL) through the exploitation of Chebyshev interpolating polynomials in each piecewise region. The goal of 2D scalability is to improve the local approximation properties of TPWL macromodels. Horizontal scalability is achieved through the reduction of number of linearization points along the trajectory; vertical scalability is obtained by extending the scope of macromodel to predict the response of a nonlinear system for inputs far from training trajectory. In this way more efficient macromodels are obtained in terms of simulation speed up of complex nonlinear systems. The methodology developed is to predict the nonlinear responses generated by faults introduced in Micro Electro-Mechanical Systems (MEMS) accelerometer during fabrication, that are used to obtain the seismic images for oil and gas discovery. We provide the implementation details and illustrate the 2D scalability concept with an example using nonlinear transmission line.

High-level modeling for operational amplifiers (opamps) has been previously carried out successfully using models generated by published automated model generation approaches. Furthermore, high-level fault modeling (HLFM) has been shown to work reasonably well using manually designed fault models. However, no evidence shows that published automated model generation approaches based on opamps have been used in HLFM. This paper describes HLFM for analog circuits using an adaptive self-tuning algorithm called multiple model generation system using delta. The generation algorithms and simulation models were written in MATLAB and the hardware description language VHDL-AMS, respectively. The properties of these self-tuning algorithms were investigated by modeling complementary metal-oxide-semiconductor opamps, and comparing simulations using the HLFM against those of the original simulation program with integrated circuit emphasis circuit utilizing transient analysis. Results show that the models can handle both linear and nonlinear fault situations with better accuracy than previously published HLFMs.

Automatic approaches for macromodel generation from SPICE-level descriptions have been of great interest over the last few years for the design of large, complex mix- signal SoCs (system-on-chips) and SiPs (system-in-packages). In this paper a novel approach termed multiple model generation system (MMGS) is developed for extracting either single-input single-output (SISO) or multiple-input single-output (MISO) macromodels from a SPICE netlist. This model generation process detects nonlinearity through variations in output error. Examples of the application of MMGS are presented for simple two-input systems incorporating a two-stage CMOS operational amplifier (op amp).