Chaoqing Tang

Chaoqing Tang, Jiaxuan Li, Huanze Zhuang et al.

To address data overload and inefficient shape-level annotation in robotic visual inspection, this paper proposes a hardware-software integrated optoelectronic architecture. A non-imaging, low-data paradigm is established to minimize annotation dependency. First, a sensor-in-the-loop strategy reconfigures a Digital Micromirror Device (DMD) as a physical optical convolutional layer, enabling photonic-domain feature extraction that unifies sensing hardware and processing software. To suppress data volume at the source, a block-based compressed sensing strategy encodes spatial information into low-dimensional temporal signals, drastically reducing redundancy. Subsequently, to bypass laborious manual defect shape annotation, natural language descriptions guide the network to align with highly generalizable features from Contrastive Language-Image Pre-training (CLIP), steering the attention maps of the optoelectronic neural network toward defect shapes. Furthermore, a Localization Accuracy for Attention (LAA) metric is proposed to quantify shape-level defect localization performance. Experiments on transparent material defect detection validate the system's effectiveness. Parametric analysis reveals how measurement matrices, compression ratios, and block sizes affect accuracy. Results show that, compared to traditional imaging, the proposed architecture maintains equivalent accuracy while reducing data volume by 90% for Vision Transformers and computational workload by 60% for Convolutional Neural Networks. This low-data paradigm offers an efficient solution for industrial automation scenarios involving massive data streams, high acquisition costs, or constrained edge resources.

Chaoqing Tang, Huanze Zhuang, Guiyun Tian et al.

Pre-trained large models attract widespread attention in recent years, but they face challenges in applications that require high interpretability or have limited resources, such as physical sensing, medical imaging, and bioinformatics. Compressed Sensing (CS) is a well-proved theory that drives many recent breakthroughs in these applications. However, as a typical under-determined linear system, CS suffers from excessively long sparse reconstruction times when using traditional iterative methods, particularly with large-scale data. Current AI methods like deep unfolding fail to substitute them because pre-trained models exhibit poor generality beyond their training conditions and dataset distributions, or lack interpretability. Instead of following the big model fervor, this paper proposes ultra-small artificial neural models called coefficients learning (CL), enabling training-free and rapid sparse reconstruction while perfectly inheriting the generality and interpretability of traditional iterative methods, bringing new feature of incorporating prior knowledges. In CL, a signal of length $n$ only needs a minimal of $n$ trainable parameters. A case study model called CLOMP is implemented for evaluation. Experiments are conducted on both synthetic and real one-dimensional and two-dimensional signals, demonstrating significant improvements in efficiency and accuracy. Compared to representative iterative methods, CLOMP improves efficiency by 100 to 1000 folds for large-scale data. Test results on eight diverse image datasets indicate that CLOMP improves structural similarity index by 292%, 98%, 45% for sampling rates of 0.1, 0.3, 0.5, respectively. We believe this method can truly usher CS reconstruction into the AI era, benefiting countless under-determined linear systems that rely on sparse solution.

Chaoqing Tang

Authentication and encryption are traditionally treated as two separate processes in wireless networks, this paper integrates user authentication into the process of solving eavesdropping attacks. A compressed sensing (CS)-based framework is proposed which manipulates the measurement matrix of CS to safeguard secure computationally. The framework is also capable of continuous authentication and transmission error correction and is robust to data loss. In detail, this paper first proposes an algorithm to generate a 2D key which depends on the physical property of communication channels. The 2D key is further used to generate authentication information and signal structure as well as encrypt original data. Then an encrypted message which contains both data and authentication information is formed for anonymous transmission. The legal receiver can split authentication information and data, and performing a data loss-robust and transmission error-robust authentication and recovery strategy. The framework is evaluated quantitatively using Monte Carlo simulation with simulated sparse signal. The secure transmission performance and authentication performance as well as data loss robustness are investigated. This framework provides an integrated security solution that efficiently safeguards the confidential, privacy and robust communication in cyber-physical systems, especially in resource-limited and safety-critical wireless networks.