ReMix: Reinforcement routing for mixtures of LoRAs in LLM finetuning

Low-rank adapters (LoRAs) are a parameter-efficient finetuning technique that injects trainable low-rank matrices into pretrained models to adapt them to new tasks. Mixture-of-LoRAs models expand neural networks efficiently by routing each layer input to a small subset of specialized LoRAs of the layer. Existing Mixture-of-LoRAs routers assign a learned routing weight to each LoRA to enable end-to-end training of the router. Despite their empirical promise, we observe that the routing weights are typically extremely imbalanced across LoRAs in practice, where only one or two LoRAs often dominate the routing weights. This essentially limits the number of effective LoRAs and thus severely hinders the expressive power of existing Mixture-of-LoRAs models. In this work, we attribute this weakness to the nature of learnable routing weights and rethink the fundamental design of the router. To address this critical issue, we propose a new router designed that we call Reinforcement Routing for Mixture-of-LoRAs (ReMix). Our key idea is using non-learnable routing weights to ensure all active LoRAs to be equally effective, with no LoRA dominating the routing weights. However, our routers cannot be trained directly via gradient descent due to our non-learnable routing weights. Hence, we further propose an unbiased gradient estimator for the router by employing the reinforce leave-one-out (RLOO) technique, where we regard the supervision loss as the reward and the router as the policy in reinforcement learning. Our gradient estimator also enables to scale up training compute to boost the predictive performance of our ReMix. Extensive experiments demonstrate that our proposed ReMix significantly outperform state-of-the-art parameter-efficient finetuning methods under a comparable number of activated parameters.

Leveraging Historical and Current Interests for Continual Sequential Recommendation

Sequential recommendation models based on the Transformer architecture show superior performance in harnessing long-range dependencies within user behavior via self-attention. However, naively updating them on continuously arriving non-stationary data streams incurs prohibitive computation costs or leads to catastrophic forgetting. To address this, we propose Continual Sequential Transformer for Recommendation (CSTRec) that effectively leverages well-preserved historical user interests while capturing current interests. At its core is Continual Sequential Attention (CSA), a linear attention mechanism that retains past knowledge without direct access to old data. CSA integrates two key components: (1) Cauchy-Schwarz Normalization that stabilizes training under uneven interaction frequencies, and (2) Collaborative Interest Enrichment that mitigates forgetting through shared, learnable interest pools. We further introduce a technique that facilitates learning for cold-start users by transferring historical knowledge from behaviorally similar existing users. Extensive experiments on three real-world datasets indicate that CSTRec outperforms state-of-the-art baselines in both knowledge retention and acquisition.

Ensuring User-side Fairness in Dynamic Recommender Systems

User-side group fairness is crucial for modern recommender systems, as it aims to alleviate performance disparity between groups of users defined by sensitive attributes such as gender, race, or age. We find that the disparity tends to persist or even increase over time. This calls for effective ways to address user-side fairness in a dynamic environment, which has been infrequently explored in the literature. However, fairness-constrained re-ranking, a typical method to ensure user-side fairness (i.e., reducing performance disparity), faces two fundamental challenges in the dynamic setting: (1) non-differentiability of the ranking-based fairness constraint, which hinders the end-to-end training paradigm, and (2) time-inefficiency, which impedes quick adaptation to changes in user preferences. In this paper, we propose FAir Dynamic rEcommender (FADE), an end-to-end framework with fine-tuning strategy to dynamically alleviate performance disparity. To tackle the above challenges, FADE uses a novel fairness loss designed to be differentiable and lightweight to fine-tune model parameters to ensure both user-side fairness and high-quality recommendations. Via extensive experiments on the real-world dataset, we empirically demonstrate that FADE effectively and efficiently reduces performance disparity, and furthermore, FADE improves overall recommendation quality over time compared to not using any new data.

Topological Augmentation for Class-Imbalanced Node Classification

Class imbalance is prevalent in real-world node classification tasks and often biases graph learning models toward majority classes. Most existing studies root from a node-centric perspective and aim to address the class imbalance in training data by node/class-wise reweighting or resampling. In this paper, we approach the source of the class-imbalance bias from an under-explored topology-centric perspective. Our investigation reveals that beyond the inherently skewed training class distribution, the graph topology also plays an important role in the formation of predictive bias: we identify two fundamental challenges, namely ambivalent and distant message-passing, that can exacerbate the bias by aggravating majority-class over-generalization and minority-class misclassification. In light of these findings, we devise a lightweight topological augmentation method ToBA to dynamically rectify the nodes influenced by ambivalent/distant message-passing during graph learning, so as to mitigate the class-imbalance bias. We highlight that ToBA is a model-agnostic, efficient, and versatile solution that can be seamlessly combined with and further boost other imbalance-handling techniques. Systematic experiments validate the superior performance of ToBA in both promoting imbalanced node classification and mitigating the prediction bias between different classes.