An Attention-Based Interaction-Aware Spatio-Temporal Graph Neural Network for Trajectory Prediction

Zhou, Hao; Ren, Dongchun; Xia, Huaxia; Fan, Mingyu; Xu, Yang; Huang, Hai

doi:10.1007/978-3-030-63823-8_5

Cited by 3 publications

(3 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…3.5.2 User Interest Representation Discrimination. Our developed contrastive paradigm not only captures the commonality among multiple behaviors but also enhances the diversity between individual nodes to mitigate oversmoothing [23,54]. During backpropagation, the gradient of the loss function L 𝑐𝑙 with respect to the parameters of RCL is computed and used to update the parameters, causing negative samples to move away from the anchor node e 𝑢 in the parameter space.…”

Section: In-depth Discussion On Rclmentioning

confidence: 99%

Multi-Relational Contrastive Learning for Recommendation

Wei,

Xia,

Huang

2023

Proceedings of the 17th ACM Conference on Recommender Systems

View full text Add to dashboard Cite

Personalized recommender systems play a crucial role in capturing users' evolving preferences over time to provide accurate and effective recommendations on various online platforms. However, many recommendation models rely on a single type of behavior learning, which limits their ability to represent the complex relationships between users and items in real-life scenarios. In such situations, users interact with items in multiple ways, including clicking, tagging as favorite, reviewing, and purchasing. To address this issue, we propose the Relation-aware Contrastive Learning (RCL) framework, which effectively models dynamic interaction heterogeneity. The RCL model incorporates a multi-relational graph encoder that captures short-term preference heterogeneity while preserving the dedicated relation semantics for different types of user-item interactions. Moreover, we design a dynamic cross-relational memory network that enables the RCL model to capture users' long-term multi-behavior preferences and the underlying evolving cross-type behavior dependencies over time. To obtain robust and informative user representations with both commonality and diversity across multi-behavior interactions, we introduce a multi-relational contrastive learning paradigm with heterogeneous shortand long-term interest modeling. Our extensive experimental studies on several real-world datasets demonstrate the superiority of the RCL recommender system over various state-of-the-art baselines in terms of recommendation accuracy and effectiveness. We provide the implementation codes for the RCL model at https://github.com/HKUDS/RCL.

show abstract

Section: In-depth Discussion On Rclmentioning

confidence: 99%

Multi-Relational Contrastive Learning for Recommendation

Wei,

Xia,

Huang

2023

Proceedings of the 17th ACM Conference on Recommender Systems

View full text Add to dashboard Cite

show abstract

“… Chen, Tian & Wu (2020) mixed the information gain with a long short-term memory network, and the improved model prediction accuracy reached 0.967. Today, LSTM is also used widely for air quality predictions, and these studies can be found in literature surveys ( Liu, Zhang & Qi, 2022 ; Bai & Shen, 2019 ; Zhou et al, 2020 ), the accuracy is up to 0.95. Yu (2020) uses one-dimensional convolutional kernels to extract features, combines LSTM model and genetic algorithm to construct the model, and the MAE is as low as 0.961.…”

Section: Introductionmentioning

confidence: 99%

Research on air quality prediction based on improved long short-term memory network algorithm

Huang

Cao

Cheng

et al. 2022

PeerJ Computer Science

View full text Add to dashboard Cite

Air quality is changing due to the influence of industry, agriculture, people’s living activities and other factors. Traditional machine learning methods generally do not consider the time series of the data itself and cannot handle long-range dependencies, thus ignoring information relevant to the predicted items and affecting the accuracy of air quality predictions. Therefore, an attention mechanism is introduced based on the long short term memory network model (LSTM), which attenuates unimportant information by controlling the proportion of the weight distribution. Finally, an integrated lightGBM+LSTM-attention model was constructed based on the light gradient boosting machine (lightGBM), and the prediction results were compared with those of 11 models. The experimental results show that the integrated model constructed in this article performs better, with the coefficient of determination (R2) of prediction accuracy reaching 0.969 and the root mean square error (RMSE) improving by 5.09, 4.94, 4.85 and 4.0 respectively compared to other models, verifying the superiority of the model.

show abstract

“…Overall, our benchmarking framework and experiments provide a rigorous setup for GNN research. Aspects of the benchmark have led to facilitating several interesting studies for GNNs such as on (i) the aggregation functions and filters [106,120,121], (ii) improving expressive power of GNNs [122][123][124], (iii) pooling mechanisms [125], (iv) graph-specific normalization and regularization [126][127][128], and (v) GNNs' robustness and efficiency [129,130] among other ideas contributed in the literature.…”

Section: Discussionmentioning

confidence: 99%

Deep learning for graph structured data

Dwivedi

View full text Add to dashboard Cite

Embarking on this PhD journey has been a remarkable experience, made possible through the invaluable support and encouragement of numerous individuals. First and foremost, my deepest gratitude goes to my supervisor, Prof. Luu Anh Tuan, whose continual guidance, motivation, and insightful feedback have been instrumental in shaping my research and my development as a scholar. I am also profoundly thankful to Prof. Xavier Bresson, who supervised my work during the initial two years of my PhD program. Under his guidance, I gained foundational knowledge in the field and learned the rigors of executing research projects. I am exceedingly fortunate to have had the opportunity to work under the guidance of both Prof. Anh Tuan and Prof. Xavier. I would also like to thank Prof. Liu Yang and Prof. Hung Dinh Nguyen for advising me as part of my thesis advisory commmittee. My parents, Mr. Shiva Shankar Dwibedi and Mrs. Lila Dwibedi, have been a constant source of inspiration, teaching me to be ambitious and resilient from an early age. Their unwavering love and support has served as my bedrock, empowering me to persevere through challenges. I extend my heartfelt thanks to my brother, sisters and brother-in-laws for their love and moral support throughout this journey. I had the privilege to intern at Sony Group Corporation in Tokyo, Japan, and Snap Inc. in Seattle, USA, which significantly enhanced my PhD experience. I would like to express my gratitude to Ishii Masato, Takuya Narihira, and Teruaki Hayashi at Sony for broadening my understanding of real-world applications. Special thanks go to my fellow interns at Sony-Siddartha, Aakash, Jai, Noel, Sri, and Christian-with whom I created lasting memories. Similarly, I am grateful for the mentorship provided by Tong Zhao, Neil Shah, and Yozen Liu at Snap Inc., as well as the journey shared with fellow colleagues-Shubham, Yushun, Clark, Zhichun, William, Ankit, Asket, and Anupam. My time in Seattle was further enriched by the friendship of Utsav, Reecha, Bidur, Suman, Sandeep, and Kriti, who made me feel at home. ix x I owe a debt of gratitude to Chaitanya Joshi, whose stimulating discussions during my early days at NTU sparked many research ideas. My colleagues Xiaoyu, Peng and Victor, who were also part of the same lab during my initial period at NTU, deserve special mention for numerous chats and their support.

show abstract

An Attention-Based Interaction-Aware Spatio-Temporal Graph Neural Network for Trajectory Prediction

Cited by 3 publications

References 17 publications

Multi-Relational Contrastive Learning for Recommendation

Multi-Relational Contrastive Learning for Recommendation

Research on air quality prediction based on improved long short-term memory network algorithm

Deep learning for graph structured data

Contact Info

Product

Resources

About