Yu Chen

Junjie Zhang, Beichen Zhang, Wenqi Sun et al.

To develop effective sequential recommender systems, numerous methods have been proposed to model historical user behaviors. Despite the effectiveness, these methods share the same fast thinking paradigm. That is, for making recommendations, these methods typically encodes user historical interactions to obtain user representations and directly match these representations with candidate item representations. However, due to the limited capacity of traditional lightweight recommendation models, this one-step inference paradigm often leads to suboptimal performance. To tackle this issue, we present a novel slow thinking recommendation model, named STREAM-Rec. Our approach is capable of analyzing historical user behavior, generating a multi-step, deliberative reasoning process, and ultimately delivering personalized recommendations. In particular, we focus on two key challenges: (1) identifying the suitable reasoning patterns in recommender systems, and (2) exploring how to effectively stimulate the reasoning capabilities of traditional recommenders. To this end, we introduce a three-stage training framework. In the first stage, the model is pretrained on large-scale user behavior data to learn behavior patterns and capture long-range dependencies. In the second stage, we design an iterative inference algorithm to annotate suitable reasoning traces by progressively refining the model predictions. This annotated data is then used to fine-tune the model. Finally, in the third stage, we apply reinforcement learning to further enhance the model generalization ability. Extensive experiments validate the effectiveness of our proposed method.

Yu Chen, Siwei Wang, Longbo Huang et al.

We study the $K$-Max combinatorial multi-armed bandits problem with continuous outcome distributions and weak value-index feedback: each base arm has an unknown continuous outcome distribution, and in each round the learning agent selects $K$ arms, obtains the maximum value sampled from these $K$ arms as reward and observes this reward together with the corresponding arm index as feedback. This setting captures critical applications in recommendation systems, distributed computing, server scheduling, etc. The continuous $K$-Max bandits introduce unique challenges, including discretization error from continuous-to-discrete conversion, non-deterministic tie-breaking under limited feedback, and biased estimation due to partial observability. Our key contribution is the computationally efficient algorithm DCK-UCB, which combines adaptive discretization with bias-corrected confidence bounds to tackle these challenges. For general continuous distributions, we prove that DCK-UCB achieves a $\widetilde{\mathcal{O}}(T^{3/4})$ regret upper bound, establishing the first sublinear regret guarantee for this setting. Furthermore, we identify an important special case with exponential distributions under full-bandit feedback. In this case, our proposed algorithm MLE-Exp enables $\widetilde{\mathcal{O}}(\sqrt{T})$ regret upper bound through maximal log-likelihood estimation, achieving near-minimax optimality.

Rui Hu, Yu Chen, Longbo Huang

Many popular practical reinforcement learning (RL) algorithms employ evolving reward functions-through techniques such as reward shaping, entropy regularization, or curriculum learning-yet their theoretical foundations remain underdeveloped. This paper provides the first finite-time convergence analysis of a single-timescale actor-critic algorithm in the presence of an evolving reward function under Markovian sampling. We consider a setting where the reward parameters may change at each time step, affecting both policy optimization and value estimation. Under standard assumptions, we derive non-asymptotic bounds for both actor and critic errors. Our result shows that an $O(1/\sqrt{T})$ convergence rate is achievable, matching the best-known rate for static rewards, provided the reward parameters evolve slowly enough. This rate is preserved when the reward is updated via a gradient-based rule with bounded gradient and on the same timescale as the actor and critic, offering a theoretical foundation for many popular RL techniques. As a secondary contribution, we introduce a novel analysis of distribution mismatch under Markovian sampling, improving the best-known rate by a factor of $\log^2T$ in the static-reward case.

Haiyun Guo, ZhiYan Hou, Yu Chen et al.

Multimodal large language models (MLLMs) require continual instruction tuning during their post-training phase to adapt to the dynamic real-world demands. However, the absence of rigorous and systematic benchmarks has hindered progress in this area. To bridge this gap, we introduce \textbf{MLLM-CTBench}, a dataset curating seven challenging tasks from six diverse domains with three contributions. First,to enable fine-grained analysis of continual learning ability, we introduce \textbf{multidimensional evaluation metrics}, which combines final answer accuracy with Chain-of-Thought (CoT) reasoning quality assessment through a carefully trained MLLM evaluator. Then, we conduct a \textbf{comprehensive evaluation of continual learning algorithms}, systematically assessing eight algorithms from four major categories to provide actionable insights for algorithm design and adoption. Finally ,we evaluate the efficacy of \textbf{Reinforcement Fine-tuning (RFT) versus Supervised Fine-tuning (SFT)} in maintaining model performance across sequential tasks during continual instruction tuning. Our experiments demonstrate that reasoning processes in MLLMs exhibit greater resilience than final outputs to forgetting during continual learning, aligning with cognitive theories of hierarchical forgetting. We further show that both model capability and task sequence significantly influence continual learning outcomes, with stronger baseline models exhibiting greater resistance to forgetting. Notably, properly regularized RFT emerges as a more robust approach than SFT for maintaining performance across tasks.One of the key contributing factors is KL-divergence regularization, without which RFT leads to even worse forgetting than SFT on old tasks though may perform better on new tasks.

Yu-Zhong Chen, Lezhi Wang, Wenxu Wang et al.

In this paper, we investigate the linear controllability framework for complex networks from a physical point of view. There are three main results. (1) If one applies control signals as determined from the structural controllability theory, there is a high probability that the control energy will diverge. Especially, if a network is deemed controllable using a single driving signal, then most likely the energy will diverge. (2) The energy required for control exhibits a power-law scaling behavior. (3) Applying additional control signals at proper nodes in the network can reduce and optimize the energy cost. We identify the fundamental structures embedded in the network, the longest control chains, which determine the control energy and give rise to the power-scaling behavior. (To our knowledge, this was not reported in any previous work on control of complex networks.) In addition, the issue of control precision is addressed. These results are supported by extensive simulations from model and real networks, physical reasoning, and mathematical analyses. Notes on the submission history of this work: This work started in late 2012. The phenomena of power-law energy scaling and energy divergence with a single controller were discovered in 2013. Strategies to reduce and optimize control energy was articulated and tested in 2013. The senior co-author (YCL) gave talks about these results at several conferences, including the NETSCI 2014 Satellite entitled "Controlling Complex Networks" on June 2, 2014. The paper was submitted to PNAS in September 2014 and was turned down. It was revised and submitted to PRX in early 2015 and was rejected. After that it was revised and submitted to Nature Communications in May 2015 and again was turned down.