Bowie Liu

Although many complex models were proposed to analyze time series data, some studies have demonstrated remarkable performance with simpler structures. A recent study proposed a non-parametric framework for 3D point cloud classification, which has the potential to be adapted for time series forecasting and enable interpretability. Inspired by the previous works, we present TSNN, a non-parametric and interpretable framework for traffic time series forecasting. TSNN consists of multiple layers that decouple the time series by matching the entries in a memory bank, where the memory bank is constructed using a similar matching process within the training set. It leverages the periodicity in traffic data to enhance forecasting accuracy while maintaining a simple model architecture. The proposed model operates without trainable parameters, preserving its inherent interpretability. In the experiments, TSNN achieves competitive performance compared to the typical deep learning models in four real-world traffic flow datasets. We also visualize the decoupling process to show the effectiveness of the components. Finally, we demonstrate the interpretability of the model and illustrate the contribution of each time step within the memory bank. Our code is available at https://github.com/pzzzzzm/TSNN_release.

Conventional deep joint source-channel coding (DJSCC) methods exhibit semantic misalignment in noisy environments due to their reliance on pixel-level metrics and inability to preserve interpretable semantic attributes, particularly under varying low signal-to-noise ratio (SNR) conditions. Semantic misalignment poses significant risks for safety-critical applications such as telemedicine and autonomous systems, where semantic misinterpretations may result in operational failures. To address these challenges, we propose a semantic-aware deep joint source-channel coding with attribute fidelity evaluation for channel-adaptive image transmission (SADJSCC), a unified framework to address semantic misalignment: semantic feature distortion, latent space displacement, and semantic undetectability. Our SADJSCC introduces a Gaussian discriminant analysis-based semantic-aware module (SAM) that generates coordinated semantic labels and semantic latent vectors through expectation-maximization optimization, resolving latent space displacement while enhancing feature dispersion. The semantic labels guide both the encoder and decoder to achieve semantic alignment. We further incorporate an attention fusion module (AFM) employing multi-head attention to dynamically weight semantic and feature latent spaces, thereby mitigating noise-induced distortions during transmission. To enable quantifiable semantic assessment independent of pixel corruption, we further develop a Semantic Attribute Fidelity Evaluation (SAFE) module that leverages Bayesian counterfactual mechanisms. Moreover, the framework integrates an SNR adaptive module (SNRAM) that dynamically recalibrates features in response to channel fluctuations. Experimental validation across CelebA, CIFAR-10, and MNIST datasets demonstrates consistent performance gains for SADJSCC. On CelebA, SADJSCC achieves an 8.4 dB peak signal-to-noise ratio (PSNR) improvement over conventional DJSCC at SNR=-4 dB, with structural similarity index measure (SSIM) enhanced by 0.16 and learned perceptual image patch similarity (LPIPS) reduced by 46%, while maintaining high computational efficiency of 12.45 dB/G. Visualizations confirm the ability to preserve semantic coherence independent of pixel-level distortions, validating its utility for safety-sensitive applications.

Dual challenges of computational efficiency and representation effectiveness exist in processing point clouds. Inspired by the TDA (Topological Data Analysis), we propose to convert the point cloud to a transposed fully connected and weighted (t-FCW) graph in order to significantly decrease the computational complexity in the following processing steps. We design a TDA pipeline called Point-FCW with a series of vectorization techniques for the 3D object point cloud feature extraction, which is plugged into the non-parametric classification head. Our experimental results demonstrate that Point-FCW achieves 75.28% accuracy on the ModelNet40 dataset with 512 points, providing a tiny, consistent, and effective representation for TDA. Furthermore, when integrated with the state-of-the-art non-parametric network Point-NN, the mixture model performs better, with an improvement of 4.47% in the OBJ-BG split of the ScanObjectNN dataset. Similarly, when integrating Point-FCW into the parametric network, PointMLP yields a performance improvement of 3.54% in the PB-T50-RS split of the ScanObjectNN dataset. The proposed Point-FCW can serve as a complementary enhancement feature when integrated into the Point-NN and PointMLP models. Moreover, the t-FCW graph representation can be efficiently converted at a rate of 3739 samples/second. Our code is available in https://github.com/hawkinglai/Point-FCW.

Accurate bus arrival time prediction is crucial for enhancing passenger experience and optimizing smart city transit systems. Existing methods, typically based on single-route, sparse stop data, struggle with the complex spatiotemporal interactions present in dense stop areas and multi-route networks, resulting in lower prediction accuracy. In this paper, we propose a frontend-enhanced time-series prediction network, in which the Multi-Relational Modeling Graph Convolution (MRMGCN) as the frontend-enhanced module, called FEN-MRMGCN. The proposed module captures spatial relationships in dense, multi-route areas by using graph convolution layers based on multi-relational modeling to aggregate spatial information. The network then uses a conventional time-series model to capture temporal dynamics. Our approach effectively combines external factors, such as traffic congestion and weather conditions, particularly in dense bus route areas, thereby significantly enhancing bus arrival time prediction accuracy. Moreover, we compile and analyze a comprehensive dataset comprising passenger flow, traffic conditions, weather information, and arrival times for both densely populated bus stop areas and regular areas in Macao. Experimental results demonstrate that our frontend-enhanced network achieves a reduction in the Mean Absolute Percentage Error (MAPE) by 15.36%, 13.44%, and 19.07% compared to traditional time series forecasting models like CNN, LSTM, and Transformer, respectively. Future research will focus on leveraging additional data sources and exploring advanced graph convolutional architectures to further elevate prediction accuracy.

Many countries across the globe face the serious issue of traffic congestion. This paper presents a low-cost graph-based traffic forecasting and congestion detection framework using online images from multiple cameras. The advantage of using a graph neural network (GNN) for traffic forecasting and detection is that it represents the traffic network in a natural way. This framework requires only images from surveillance cameras without any other sensors. It converts the online images into two types of data: traffic volume and image-based traffic occupancy. A clustering-based graph construction method is proposed to build a graph based on the traffic network. For traffic forecasting, multiple models, including statistical models and deep graph convolutional neural networks (GCNs), are used and compared using the extracted data. The framework uses logistic regression to determine the threshold of traffic congestion. In the experiment, we found that the Decoupled Dynamic Spatial-Temporal Graph Neural Network (D2STGNN) model achieved the best performance on the collected dataset. We also propose a threshold-based method for detecting traffic congestion using traffic volume and image-based traffic occupancy. This framework provides a low-cost solution for traffic forecasting and congestion detection when only surveillance images are available.

The concept of the “smart city” has emerged with the advancement of technology, but some facilities are not sufficiently intelligent, such as parking lots. Hence, this paper proposes an inexpensive and plug-to-play camera-based smart parking system for airports. The system utilizes inverse perspective mapping (IPM) to provide an aerial view image of the parking lot, which is then processed to extract parking space information. The system also includes a guidance system to assist drivers in finding available parking spaces. The system is simulated on a 3D scene based on the parking lot of Macao International Airport. In the experiment, our system achieved an accuracy rate of 97.03% and a mean distance error of 8.59 pixels. This research study shows the potential of enhancing parking lots using only cameras as data collectors, and the results show that the system is capable of providing accurate and useful information. It performs well in parking lots with open space, in particular. Moreover, it is an economical solution for implementing a smart parking lot.