Jianguo Li

Zelin Ni, Hang Yu, Shizhan Liu et al.

Bases have become an integral part of modern deep learning-based models for time series forecasting due to their ability to act as feature extractors or future references. To be effective, a basis must be tailored to the specific set of time series data and exhibit distinct correlation with each time series within the set. However, current state-of-the-art methods are limited in their ability to satisfy both of these requirements simultaneously. To address this challenge, we propose BasisFormer, an end-to-end time series forecasting architecture that leverages learnable and interpretable bases. This architecture comprises three components: First, we acquire bases through adaptive self-supervised learning, which treats the historical and future sections of the time series as two distinct views and employs contrastive learning. Next, we design a Coef module that calculates the similarity coefficients between the time series and bases in the historical view via bidirectional cross-attention. Finally, we present a Forecast module that selects and consolidates the bases in the future view based on the similarity coefficients, resulting in accurate future predictions. Through extensive experiments on six datasets, we demonstrate that BasisFormer outperforms previous state-of-the-art methods by 11.04\% and 15.78\% respectively for univariate and multivariate forecasting tasks. Code is available at: \url{https://github.com/nzl5116190/Basisformer}

Shiyi Zhu, Jing Ye, Wei Jiang et al.

Self-attention and position embedding are two key modules in transformer-based Large Language Models (LLMs). However, the potential relationship between them is far from well studied, especially for long context window extending. In fact, anomalous behaviors harming long context extrapolation exist between Rotary Position Embedding (RoPE) and vanilla self-attention unveiled by our work. To address this issue, we propose a novel attention mechanism, CoCA (Collinear Constrained Attention). Specifically, we enforce a collinear constraint between $Q$ and $K$ to seamlessly integrate RoPE and self-attention. While only adding minimal computational and spatial complexity, this integration significantly enhances long context window extrapolation ability. We provide an optimized implementation, making it a drop-in replacement for any existing transformer-based models. Extensive experiments show that CoCA performs extraordinarily well in extending context windows. A CoCA-based GPT model, trained with a context length of 512, can seamlessly extend the context window up to 32K (60$\times$), without any fine-tuning. Additionally, by dropping CoCA in LLaMA-7B, we achieve extrapolation up to 32K within only 2K training length. Our code is publicly available at: https://github.com/codefuse-ai/Collinear-Constrained-Attention

Jianguo Li, Yong Tang, Jiemin Chen

Recommender systems (RSs) have been a widely exploited approach to solving the information overload problem. However, the performance is still limited due to the extreme sparsity of the rating data. With the popularity of Web 2.0, the social tagging system provides more external information to improve recommendation accuracy. Although some existing approaches combine the matrix factorization models with co-occurrence properties and context of tags, they neglect the issue of tag sparsity without the commonly associated tags problem that would also result in inaccurate recommendations. Consequently, in this paper, we propose a novel hybrid collaborative filtering model named WUDiff_RMF, which improves Regularized Matrix Factorization (RMF) model by integrating Weighted User-Diffusion-based CF algorithm(WUDiff) that obtains the information of similar users from the weighted tripartite user-item-tag graph. This model aims to capture the degree correlation of the user-item-tag tripartite network to enhance the performance of recommendation. Experiments conducted on four real-world datasets demonstrate that our approach significantly performs better than already widely used methods in the accuracy of recommendation. Moreover, results show that WUDiff_RMF can alleviate the data sparsity, especially in the circumstance that users have made few ratings and few tags.