Egocentric video-based models capture rich semantic information and have demonstrated strong performance in human activity recognition (HAR). However, their high power consumption, privacy concerns, and dependence on lighting conditions limit their feasibility for continuous on-device recognition. In contrast, inertial measurement unit (IMU) sensors offer an energy-efficient and privacy-preserving alternative, yet they suffer from limited large-scale annotated datasets, leading to weaker generalization in downstream tasks. To bridge this gap, we propose COMODO, a cross-modal self-supervised distillation framework that transfers rich semantic knowledge from the video modality to the IMU modality without requiring labeled annotations. COMODO leverages a pretrained and frozen video encoder to construct a dynamic instance queue, aligning the feature distributions of video and IMU embeddings. By distilling knowledge from video representations, our approach enables the IMU encoder to inherit rich semantic information from video while preserving its efficiency for real-world applications. Experiments on multiple egocentric HAR datasets demonstrate that COMODO consistently improves downstream classification performance, achieving results comparable to or exceeding fully supervised fine-tuned models. Moreover, COMODO exhibits strong cross-dataset generalization. Benefiting from its simplicity, our method is also generally applicable to various video and time-series pre-trained models, offering the potential to leverage more powerful teacher and student foundation models in future research. The code is available at https://github.com/Breezelled/COMODO .

In this work, we bridge the gap between wearable sensor technology and personalized AI assistants by enabling Large Language Models (LLMs) to understand time-series tasks like human activity recognition (HAR). Despite the strong reasoning and generalization capabilities of LLMs, leveraging them for sensor data tasks remains largely unexplored. This gap stems from challenges like the lack of semantic context in time-series data, computational limitations, and LLMs' difficulty processing numerical inputs. To address these issues, we introduce SensorLLM, a two-stage framework to unlock LLMs' potential for sensor data tasks. In the Sensor-Language Alignment Stage, we introduce special tokens for each sensor channel and automatically generate trend-descriptive text to align sensor data with textual inputs, enabling SensorLLM to capture numerical changes, channel-specific information, and sensor data of varying lengths-capabilities that existing LLMs typically struggle with, all without the need for human annotations. Next, in Task-Aware Tuning Stage, we refine the model for HAR classification using the frozen LLM and alignment module, achieving performance on par with or surpassing state-of-the-art models. We further demonstrate that SensorLLM evolves into an effective sensor learner, reasoner, and classifier through Sensor-Language Alignment, enabling it to generalize across diverse datasets for HAR tasks. We strongly believe our work lays the stepstone for future time-series and text alignment research, offering a path toward foundation models for sensor data.

Training models on spatio-temporal (ST) data poses an open problem due to the complicated and diverse nature of the data itself, and it is challenging to ensure the model's performance directly trained on the original ST data. While limiting the variety of training data can make training easier, it can also lead to a lack of knowledge and information for the model, resulting in a decrease in performance. To address this challenge, we presented an innovative paradigm that incorporates three separate forms of curriculum learning specifically targeting from spatial, temporal, and quantile perspectives. Furthermore, our framework incorporates a stacking fusion module to combine diverse information from three types of curriculum learning, resulting in a strong and thorough learning process. We demonstrated the effectiveness of this framework with extensive empirical evaluations, highlighting its better performance in addressing complex ST challenges. We provided thorough ablation studies to investigate the effectiveness of our curriculum and to explain how it contributes to the improvement of learning efficiency on ST data.

Recently, learning urban region representations utilizing multi-modal data (information views) has become increasingly popular, for deep understanding of the distributions of various socioeconomic features in cities. However, previous methods usually blend multi-view information in a posteriors stage, falling short in learning coherent and consistent representations across different views. In this paper, we form a new pipeline to learn consistent representations across varying views, and propose the multi-view Contrastive Prediction model for urban Region embedding (ReCP), which leverages the multiple information views from point-of-interest (POI) and human mobility data. Specifically, ReCP comprises two major modules, namely an intra-view learning module utilizing contrastive learning and feature reconstruction to capture the unique information from each single view, and inter-view learning module that perceives the consistency between the two views using a contrastive prediction learning scheme. We conduct thorough experiments on two downstream tasks to assess the proposed model, i.e., land use clustering and region popularity prediction. The experimental results demonstrate that our model outperforms state-of-the-art baseline methods significantly in urban region representation learning.