Chia-Yuan Chang, Zhimeng Jiang, Vineeth Rakesh, Menghai Pan, Chin-Chia Michael Yeh, Guanchu Wang, Mingzhi Hu, Zhichao Xu, Yan Zheng, Mahashweta Das, Na Zou. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

This paper introduces a novel approach to fine-tuning transformer-based models for various spatial-temporal downstream tasks. While fine-tuning approaches have shown remarkable success in fields like natural language processing, their efficacy in human-generated spatial-temporal data is often hindered by the complex spatial-temporal correlation and extensive reliance on labeled data. We introduce Knowledge Graph-Guided Spatial-Temporal Cross-task Fine-Tuning method, i.e., KG-STFT, a cross-task fine-tuning approach designed for adapting to data-scarce spatial-temporal tasks by leveraging rich knowledge embedded in similar downstream tasks and pre-trained model. Our KG-STFT framework utilizes i) a non-linear knowledge ensembler to capture and integrate knowledge embedded in different transformer blocks, and ii) constructs a task knowledge graph to "transfer" knowledge from data-rich to data-scarce tasks. Empirical experiments on real-world taxi trajectory data show that KG-STFT outperforms baselines, especially in data-scarce tasks, by leveraging task commonalities to improve fine-tuning.

Adversarial steganography presents a significant challenge in digital media security, leveraging adversarial perturbations to obscure steganographic patterns and evade detection by traditional steganalysis methods. This paper proposes a universal detection method based on Directional Adversarial Noise Search (DANS) to improve the robustness of steganalysis against adversarial steganography. Specifically, an augmentation network is employed to generate adversarial noise distributions, guided by a novel DANS loss function that enables precise optimization of noise distribution. A threshold constraint is further applied to ensure the controllability of the generated noise. By embedding the generated perturbations into both cover and stego images, a robust augmented dataset is constructed to improve detection performance. Experimental results demonstrate that the proposed method improves detection accuracy across various adversarial steganography techniques, achieving up to a 17.14% increase. Moreover, it exhibits superior generalization ability in cross-dataset evaluations, highlighting its effectiveness and robustness in diverse scenarios.

Coverless generative steganography is a highly secure method of information hiding. With the advent of the AI-generated content (AIGC) era, the widespread dissemination of generative content on the internet provides an excellent hiding environment for generative steganographic images. Generative steganographic images do not require the participation of carrier images, making existing steganalysis methods expired. However, there are currently no detection methods specifically targeting generative steganographic content. To address this gap, we propose a steganalysis method for generative steganographic images. Our approach focuses on the intrinsic differences between generative steganographic images and ordinary generative images. Through comparative analysis, we propose optimizing the detection model using mutual information estimation. We hypothesize about the distribution characteristics of steganographic signals and design a feature discrimination loss function to further guide the model’s optimization. In addition to designing a feature extraction network to extract features from different image regions, we also incorporate an image classification model pretrained on a large dataset to extract classification features for the final classification. Experimental results in various training and testing scenarios demonstrate that the proposed model not only possesses excellent detection capability but also exhibits reliable generalization compared to other models. Furthermore, we provide necessary descriptions and analysis to validate the rationale behind the network design.

Human-generated Spatial-Temporal Data (HSTD), represented as trajectory sequences, has undergone a data revolution, thanks to advances in mobile sensing, data mining, and AI. Previous studies have revealed the effectiveness of employing attention mechanisms to analyze massive HSTD. However, traditional attention models face challenges when managing lengthy and noisy trajectories as their computation comes with large memory overheads. Furthermore, attention scores within HSTD trajectories are sparse (i.e., most of the scores are zeros), and clustered with varying lengths (i.e., consecutive tokens clustered with similar scores). To address these challenges, we introduce an innovative strategy named Memory-efficient Trajectory Attention (MeTA). We leverage complicated spatial-temporal features (e.g., traffic speed, proximity to PoIs) and design an innovative feature-based trajectory partition technique to shrink trajectory length. Additionally, we present a learnable dynamic sorting mechanism, with which attention is only computed between sub-trajectories that have prominent correlations. Empirical validations using real-world HSTD demonstrate that our approach not only yields competitive results but also significantly lowers memory usage compared with state-of-the-art methods. Our approach presents innovative solutions for memory-efficient trajectory attention, offering valuable insights for handling HSTD efficiently.

Advancements in mobile sensing, data mining, and artificial intelligence have revolutionized the collection and analysis of Human-generated Spatial-Temporal Data (HSTD), paving the way for diverse applications across multiple domains. However, previous works have primarily focused on designing task-specific models for different problems, which lack transferability and generalizability when confronted with diverse HSTD. Additionally, these models often require a large amount of labeled data for optimal performance. While pre-trained models in Natural Language Processing (NLP) and Computer Vision (CV) domains have showcased impressive transferability and generalizability, similar efforts in the spatial-temporal data domain have been limited. In this paper, we take the lead and introduce the Spatial-Temporal Pre-Training model, $i.e$., STPT, which is connected with a self-supervised learning task, to address these limitations. STPT enables the creation of robust and versatile representations of HSTD. We validate our framework using real-world data and demonstrate its efficacy through two downstream tasks, $i.e$., trajectory classification and driving activity identification $(e.g$., identifying seeking $vs$. serving behaviors in taxi trajectories). Our results achieve an accuracy of 83.125% (16.2% higher than the average baseline) for human mobility identification and an accuracy of 77.88% (13.0% higher than the average baseline) for the human activity identification task. These outcomes underscore the potential of our pre-trained model for diverse downstream applications within the spatial-temporal data domain.

The Human Mobility Signature Identification (HuMID) problem aims at determining whether the incoming trajectories were generated by a claimed agent from the historical movement trajectories of a set of individual human agents such as pedestrians and taxi drivers. The HuMID problem is significant, and its solutions have a wide range of real-world applications, such as criminal identification for police departments, risk assessment for auto insurance providers, driver verification in ride-sharing services, and so on. Though Deep neural networks (DNN) based HuMID models on spatial-temporal mobility fingerprint similarity demonstrate remarkable performance in effectively identifying human agents' mobility signatures, it is vulnerable to adversarial attacks as other DNN-based models. Therefore, in this paper, we propose a Spatial-Temporal iterative Fast Gradient Sign Method with L0 regularization - ST-iFGSM - to detect the vulnerability and enhance the robustness of HuMID models. Extensive experiments with real-world taxi trajectory data demonstrate the efficiency and effectiveness of our ST-iFGSM algorithm. We tested our method on both the ST-SiameseNet and an LSTM-based HuMID classification model. It shows that ST-iFGSM can generate successful attacks to fool the HuMID models with only a few steps of attack in a small portion of the trajectories. The generated attacks can be used as augmented data to update and improve the HuMID model accuracy significantly from 47.36% to 76.18% on testing samples after the attack(86.25% on the original testing samples).

Convolutional Neural Networks (CNNs) have made remarkable progress in steganalysis. However, they struggle to detect adversarial steganography accurately which merges adversarial samples and steganography. While handcrafted models show limited vulnerability to adversarial steganography, their accuracy pales in comparison to that of CNN analyzers. To address these limitations head-on, we propose TStegNet, an innovative two-stream CNN steganalyzer designed to detect adversarial steganography. TStegNet leverages confidence artifacts and pixel artifacts, enabling a comprehensive analysis of hidden information. Specifically, we design a confidence loss function and apply backpropagation to amplify the confidence artifacts, which enhances the performance of our model. Additionally, we use the feature similarity function to minimize the impact of adversarial perturbation. Extensive experiments reveal that proposed TStegNet outperforms existing state-of-the-art methods, representing a significant milestone in the fight against adversarial steganography.