Dawei Feng

Ruiqing Sun, Sen Yang, Dawei Feng et al.

Offline multi-objective optimization (Offline MOO) aims to discover novel Pareto-optimal designs based on static datasets without expensive environment interactions. While recent generative methods have achieved notable success, they predominantly rely on external surrogate models. This dependency introduces significant computational overhead, suffers from deceptive evaluations, and deviates from the prevailing paradigm of jointly training mainstream generative models with conditions. To address these bottlenecks, we propose ParetoPilot, a novel zero-surrogate diffusion framework for offline MOO. ParetoPilot fully leverages the conditional priors inherently embedded within pre-trained diffusion models. At its core, the framework introduces the Infer-Perturb-Guide (IPG) engine, which is seamlessly interleaved within the unconditional denoising steps of the reverse generation process. First, it implicitly infers the instantaneous objective direction by matching conditional and unconditional noise predictions. Next, it mathematically orthogonalizes a parallel gravity field for strict convergence and an edgeness-aware repulsive force for mutual diversity, creating a dynamically annealed perturbation vector. Finally, this perturbed target seamlessly steers the generation process via standard Classifier-Free Guidance (CFG). Extensive experiments across 51 tasks demonstrate that ParetoPilot outperforms 14 state-of-the-art surrogate-based and inverse generative baselines. By eliminating auxiliary proxy training, our approach preserves data privacy while achieving hypervolume improvement and robust Pareto front coverage.

Chao Chen, Zijian Gao, Kele Xu et al.

To handle the sparsity of the extrinsic rewards in reinforcement learning, researchers have proposed intrinsic reward which enables the agent to learn the skills that might come in handy for pursuing the rewards in the future, such as encouraging the agent to visit novel states. However, the intrinsic reward can be noisy due to the undesirable environment's stochasticity and directly applying the noisy value predictions to supervise the policy is detrimental to improve the learning performance and efficiency. Moreover, many previous studies employ $\ell^2$ norm or variance to measure the exploration novelty, which will amplify the noise due to the square operation. In this paper, we address aforementioned challenges by proposing a novel curiosity leveraging the nuclear norm maximization (NNM), which can quantify the novelty of exploring the environment more accurately while providing high-tolerance to the noise and outliers. We conduct extensive experiments across a variety of benchmark environments and the results suggest that NNM can provide state-of-the-art performance compared with previous curiosity methods. On 26 Atari games subset, when trained with only intrinsic reward, NNM achieves a human-normalized score of 1.09, which doubles that of competitive intrinsic rewards-based approaches. Our code will be released publicly to enhance the reproducibility.

Zijian Gao, Kele Xu, Yuanzhao Zhai et al.

Under sparse extrinsic reward settings, reinforcement learning has remained challenging, despite surging interests in this field. Previous attempts suggest that intrinsic reward can alleviate the issue caused by sparsity. In this article, we present a novel intrinsic reward that is inspired by human learning, as humans evaluate curiosity by comparing current observations with historical knowledge. Our method involves training a self-supervised prediction model, saving snapshots of the model parameters, and using nuclear norm to evaluate the temporal inconsistency between the predictions of different snapshots as intrinsic rewards. We also propose a variational weighting mechanism to assign weight to different snapshots in an adaptive manner. Our experimental results on various benchmark environments demonstrate the efficacy of our method, which outperforms other intrinsic reward-based methods without additional training costs and with higher noise tolerance. This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible.

Ruiqing Sun, Dawei Feng, Sen Yang et al.

In 3D molecular discovery, optimizing conflicting physicochemical properties while strictly adhering to complex structural constraints constitutes a Constrained Multi-Objective Optimization Problem (CMOP). Solving this remains highly challenging: applying traditional Evolutionary Algorithm (EA) operators directly to 3D coordinates destroys chemical validity, whereas valid 3D diffusion models act as rigid generators unable to adapt to novel objectives without retraining. Moreover, employing traditional EA frameworks causes a severe loss of structural diversity, ultimately impairing algorithmic convergence. To overcome these challenges, we propose the Evolutionary-Guided Diffusion (EGD) operator, which executes crossover and mutation exclusively within the continuous noise space at an appropriate noise intensity. EGD enables topological hybridization while leveraging a pre-trained denoising network to project intermediate states back onto the valid chemical manifold. To tackle Multi-Objective Problems (MOPs), we introduce a Structure-Aware Environmental Selection (SAES) mechanism that explicitly enforces geometric diversity. Building upon this, to specifically solve CMOPs, we develop the Diffusion-based Evolutionary Molecular Optimization (DEMO) framework, utilizing a tri-population architecture with distinct responsibilities to safely navigate disjoint feasible regions. Extensive experiments across single-property targeting, unconstrained MOPs, multi-fragment constrained generation, and 3D protein-ligand docking demonstrate that DEMO comprehensively outperforms train-free guidance methods and EA baselines. Without any model retraining, DEMO successfully discovers highly diverse, chemically valid Pareto frontiers, establishing a robust paradigm for complex 3D molecular optimization.

Zijian Gao, YiYing Li, Kele Xu et al.

The sparsity of extrinsic rewards poses a serious challenge for reinforcement learning (RL). Currently, many efforts have been made on curiosity which can provide a representative intrinsic reward for effective exploration. However, the challenge is still far from being solved. In this paper, we present a novel curiosity for RL, named DyMeCu, which stands for Dynamic Memory-based Curiosity. Inspired by human curiosity and information theory, DyMeCu consists of a dynamic memory and dual online learners. The curiosity arouses if memorized information can not deal with the current state, and the information gap between dual learners can be formulated as the intrinsic reward for agents, and then such state information can be consolidated into the dynamic memory. Compared with previous curiosity methods, DyMeCu can better mimic human curiosity with dynamic memory, and the memory module can be dynamically grown based on a bootstrap paradigm with dual learners. On multiple benchmarks including DeepMind Control Suite and Atari Suite, large-scale empirical experiments are conducted and the results demonstrate that DyMeCu outperforms competitive curiosity-based methods with or without extrinsic rewards. We will release the code to enhance reproducibility.

Xu Zhang, Xudong Gong, Jiacheng Qin et al.

Current evaluations of large language models aggregate performance across diverse tasks into single scores. This obscures fine-grained ability variation, limiting targeted model improvement and ability-guided selection for specific tasks. Motivated by this gap, we propose a cognitive diagnostic framework that estimates model abilities across multiple fine-grained dimensions. For mathematics, we construct a 35-dimensional ability taxonomy grounded in cognitive theory and domain knowledge. The framework employs multidimensional Item Response Theory with an item-ability association matrix to estimate fine-grained ability levels, which in turn enable prediction of performance on unseen items (questions of benchmark). Evaluated on 41 models, our approach demonstrates strong criterion validity, consistent ability estimates across benchmarks, and accurate prediction of unseen items with AUC ranging from 0.80 to 0.89 within benchmarks and from 0.77 to 0.86 across benchmarks, substantially exceeding trivial baselines. The framework generalizes across scientific domains, producing consistent diagnostic performance in physics (27 dimensions), chemistry (58 dimensions), and computer science (12 dimensions). This work establishes a principled framework for fine-grained assessment of abilities, with potential applications in targeted training, ability-guided model selection, and ability-aware benchmark design.

Zhishan Zhao, Sen Yang, Guohui Liu et al.

As a critical component for online advertising and marking, click-through rate (CTR) prediction has draw lots of attentions from both industry and academia field. Recently, the deep learning has become the mainstream methodological choice for CTR. Despite of sustainable efforts have been made, existing approaches still pose several challenges. On the one hand, high-order interaction between the features is under-explored. On the other hand, high-order interactions may neglect the semantic information from the low-order fields. In this paper, we proposed a novel prediction method, named FINT, that employs the Field-aware INTeraction layer which captures high-order feature interactions while retaining the low-order field information. To empirically investigate the effectiveness and robustness of the FINT, we perform extensive experiments on the three realistic databases: KDD2012, Criteo and Avazu. The obtained results demonstrate that the FINT can significantly improve the performance compared to the existing methods, without increasing the amount of computation required. Moreover, the proposed method brought about 2.72\% increase to the advertising revenue of a big online video app through A/B testing. To better promote the research in CTR field, we released our code as well as reference implementation at: https://github.com/zhishan01/FINT.

Yuanzhao Zhai, Han Zhang, Yu Lei et al.

Reinforcement learning from human feedback (RLHF) emerges as a promising paradigm for aligning large language models (LLMs). However, a notable challenge in RLHF is overoptimization, where beyond a certain threshold, the pursuit of higher rewards leads to a decline in human preferences. In this paper, we observe the weakness of KL regularization which is commonly employed in existing RLHF methods to address overoptimization. To mitigate this limitation, we scrutinize the RLHF objective in the offline dataset and propose uncertainty-penalized RLHF (UP-RLHF), which incorporates uncertainty regularization during RL-finetuning. To enhance the uncertainty quantification abilities for reward models, we first propose a diverse low-rank adaptation (LoRA) ensemble by maximizing the nuclear norm of LoRA matrix concatenations. Then we optimize policy models utilizing penalized rewards, determined by both rewards and uncertainties provided by the diverse reward LoRA ensembles. Our experimental results, based on two real human preference datasets, showcase the effectiveness of diverse reward LoRA ensembles in quantifying reward uncertainty. Additionally, uncertainty regularization in UP-RLHF proves to be pivotal in mitigating overoptimization, thereby contributing to the overall performance.

Yuanzhao Zhai, Yiying Li, Zijian Gao et al.

Model-based offline reinforcement learning (RL) has made remarkable progress, offering a promising avenue for improving generalization with synthetic model rollouts. Existing works primarily focus on incorporating pessimism for policy optimization, usually via constructing a Pessimistic Markov Decision Process (P-MDP). However, the P-MDP discourages the policies from learning in out-of-distribution (OOD) regions beyond the support of offline datasets, which can under-utilize the generalization ability of dynamics models. In contrast, we propose constructing an Optimistic MDP (O-MDP). We initially observed the potential benefits of optimism brought by encouraging more OOD rollouts. Motivated by this observation, we present ORPO, a simple yet effective model-based offline RL framework. ORPO generates Optimistic model Rollouts for Pessimistic offline policy Optimization. Specifically, we train an optimistic rollout policy in the O-MDP to sample more OOD model rollouts. Then we relabel the sampled state-action pairs with penalized rewards and optimize the output policy in the P-MDP. Theoretically, we demonstrate that the performance of policies trained with ORPO can be lower-bounded in linear MDPs. Experimental results show that our framework significantly outperforms P-MDP baselines by a margin of 30%, achieving state-of-the-art performance on the widely-used benchmark. Moreover, ORPO exhibits notable advantages in problems that require generalization.

Ruiqing Sun, Dawei Feng, Sheng Qi et al.

Existing evolutionary algorithms for Constrained Multi-objective Optimization Problems (CMOPs) typically treat all constraints uniformly, overlooking their distinct geometric relationships with the true Constrained Pareto Front (CPF). In reality, constraints play different roles: some directly shape the final CPF, some create infeasible obstacles, while others are irrelevant. To exploit this insight, we propose a novel algorithm named RCCMO, which sequentially performs unconstrained exploration, single-constraint exploitation, and full-constraint refinement. The core innovation of RCCMO lies in a constraint prioritization method derived from these geometric insights, seamlessly coupled with a unique dual-directional search mechanism. Specifically, RCCMO first prioritizes constraints that constitute the final CPF, approaching them from the evolutionary direction (optimizing objectives) to locate the CPF directly shaped by single-constraint boundaries. Subsequently, for constraints that merely hinder the population's progress, RCCMO searches from the anti-evolutionary direction (targeting the infeasible boundaries where hindering constraints intersect with the CPF) to effectively discover how these constraints obstruct and form the final CPF. Meanwhile, irrelevant constraints are intentionally bypassed. Furthermore, a series of specialized mechanisms are proposed to accelerate the algorithm's execution, reduce heuristic misjudgments, and dynamically adjust search directions in real time. Extensive experiments on 5 benchmark test suites and 29 real-world CMOPs demonstrate that RCCMO significantly outperforms seven state-of-the-art algorithms.

Dawei Feng, Di Mei, Huiri Tan et al.

Large Language Models (LLMs) have shown remarkable proficiency in natural language understanding (NLU), opening doors for innovative applications. We introduce StreamLink - an LLM-driven distributed data system designed to improve the efficiency and accessibility of data engineering tasks. We build StreamLink on top of distributed frameworks such as Apache Spark and Hadoop to handle large data at scale. One of the important design philosophies of StreamLink is to respect user data privacy by utilizing local fine-tuned LLMs instead of a public AI service like ChatGPT. With help from domain-adapted LLMs, we can improve our system's understanding of natural language queries from users in various scenarios and simplify the procedure of generating database queries like the Structured Query Language (SQL) for information processing. We also incorporate LLM-based syntax and security checkers to guarantee the reliability and safety of each generated query. StreamLink illustrates the potential of merging generative LLMs with distributed data processing for comprehensive and user-centric data engineering. With this architecture, we allow users to interact with complex database systems at different scales in a user-friendly and security-ensured manner, where the SQL generation reaches over 10\% of execution accuracy compared to baseline methods, and allow users to find the most concerned item from hundreds of millions of items within a few seconds using natural language.

Qiang Wang, Dawei Feng, Xu Zhang et al.

Instruction tuning has emerged as a paramount method for tailoring the behaviors of LLMs. Recent work has unveiled the potential for LLMs to achieve high performance through fine-tuning with a limited quantity of high-quality instruction data. Building upon this approach, we further explore the impact of prompt's robustness on the selection of high-quality instruction data. This paper proposes a pioneering framework of high-quality online instruction data mining for instruction tuning, focusing on the impact of prompt's robustness on the data mining process. Our notable innovation, is to generate the adversarial instruction data by conducting the attack for the prompt of online instruction data. Then, we introduce an Adversarial Instruction-Following Difficulty metric to measure how much help the adversarial instruction data can provide to the generation of the corresponding response. Apart from it, we propose a novel Adversarial Instruction Output Embedding Consistency approach to select high-quality online instruction data. We conduct extensive experiments on two benchmark datasets to assess the performance. The experimental results serve to underscore the effectiveness of our proposed two methods. Moreover, the results underscore the critical practical significance of considering prompt's robustness.

Zheng Luo, Ming Feng, Zijian Gao et al.

The emergence of deep learning (DL) has provided great opportunities for the high-throughput analysis of atomic-resolution micrographs. However, the DL models trained by image patches in fixed size generally lack efficiency and flexibility when processing micrographs containing diversified atomic configurations. Herein, inspired by the similarity between the atomic structures and graphs, we describe a few-shot learning framework based on an equivariant graph neural network (EGNN) to analyze a library of atomic structures (e.g., vacancies, phases, grain boundaries, doping, etc.), showing significantly promoted robustness and three orders of magnitude reduced computing parameters compared to the image-driven DL models, which is especially evident for those aggregated vacancy lines with flexible lattice distortion. Besides, the intuitiveness of graphs enables quantitative and straightforward extraction of the atomic-scale structural features in batches, thus statistically unveiling the self-assembly dynamics of vacancy lines under electron beam irradiation. A versatile model toolkit is established by integrating EGNN sub-models for single structure recognition to process images involving varied configurations in the form of a task chain, leading to the discovery of novel doping configurations with superior electrocatalytic properties for hydrogen evolution reactions. This work provides a powerful tool to explore structure diversity in a fast, accurate, and intelligent manner.

Yuanzhao Zhai, Zhuo Zhang, Kele Xu et al.

Aligning with human preference datasets has been critical to the success of large language models (LLMs). Reinforcement learning from human feedback (RLHF) employs a costly reward model to provide feedback for on-policy sampling responses. Recently, offline methods that directly fit responses with binary preferences in the dataset have emerged as alternatives. However, existing methods do not explicitly model preference strength information, which is crucial for distinguishing different response pairs. To overcome this limitation, we propose Online Self-Preferring (OSP) language models to learn from self-generated response pairs and self-judged preference strengths. For each prompt and corresponding self-generated responses, we introduce a ranked pairing method to construct multiple response pairs with preference strength information. We then propose the soft-preference cross-entropy loss to leverage such information. Empirically, we demonstrate that leveraging preference strength is crucial for avoiding overfitting and enhancing alignment performance. OSP achieves state-of-the-art alignment performance across various metrics in two widely used human preference datasets. OSP is parameter-efficient and more robust than the dominant online method, RLHF when limited offline data are available and generalizing to out-of-domain tasks. Moreover, OSP language models established by LLMs with proficiency in self-preferring can efficiently self-improve without external supervision.

Dawei Feng, Yihai Zhang, Zhixuan Xu

Pretrained Large Language Models (LLM) such as ChatGPT, Claude, etc. have demonstrated strong capabilities in various fields of natural language generation. However, there are still many problems when using LLM in specialized domain-specific fields. When using generative AI to process downstream tasks, a common approach is to add new knowledge (e.g., private domain knowledge, cutting-edge information) to a pretrained model through continued training or fine-tuning. However, whether there is a universal paradigm for domain adaptation training is still an open question. In this article, we proposed Information Gain Optimized Tokenizer (IGOT), which analyzes the special token set of downstream tasks, constructs a new subset using heuristic function $φ$ with the special token and its information gain, to build new domain-specific tokenizer, and continues pretraining on the downstream task data. We explored the many positive effects of this method's customized tokenizer on domain-adaptive pretraining and verified this method can perform better than the ordinary method of just collecting data and fine-tuning. Based on our experiment, the continued pretraining process of IGOT with LLaMA-7B achieved 11.9\% token saving, 12.2\% training time saving, and 5.8\% maximum GPU VRAM usage saving, combined with the T5 model, we can even reach a 31.5\% of training time saving, making porting general generative AI to specific domains more effective than before. In domain-specific tasks, supervised $IGOT_τ$ shows great performance on reducing both the convergence radius and convergence point during keep pretraining.

Kele Xu, Haibo Mi, Dawei Feng et al.

Valuable training data is often owned by independent organizations and located in multiple data centers. Most deep learning approaches require to centralize the multi-datacenter data for performance purpose. In practice, however, it is often infeasible to transfer all data to a centralized data center due to not only bandwidth limitation but also the constraints of privacy regulations. Model averaging is a conventional choice for data parallelized training, but its ineffectiveness is claimed by previous studies as deep neural networks are often non-convex. In this paper, we argue that model averaging can be effective in the decentralized environment by using two strategies, namely, the cyclical learning rate and the increased number of epochs for local model training. With the two strategies, we show that model averaging can provide competitive performance in the decentralized mode compared to the data-centralized one. In a practical environment with multiple data centers, we conduct extensive experiments using state-of-the-art deep network architectures on different types of data. Results demonstrate the effectiveness and robustness of the proposed method.

Dawei Feng, Kele Xu, Haibo Mi et al.

Acoustic scene classification is an intricate problem for a machine. As an emerging field of research, deep Convolutional Neural Networks (CNN) achieve convincing results. In this paper, we explore the use of multi-scale Dense connected convolutional neural network (DenseNet) for the classification task, with the goal to improve the classification performance as multi-scale features can be extracted from the time-frequency representation of the audio signal. On the other hand, most of previous CNN-based audio scene classification approaches aim to improve the classification accuracy, by employing different regularization techniques, such as the dropout of hidden units and data augmentation, to reduce overfitting. It is widely known that outliers in the training set have a high negative influence on the trained model, and culling the outliers may improve the classification performance, while it is often under-explored in previous studies. In this paper, inspired by the silence removal in the speech signal processing, a novel sample dropout approach is proposed, which aims to remove outliers in the training dataset. Using the DCASE 2017 audio scene classification datasets, the experimental results demonstrates the proposed multi-scale DenseNet providing a superior performance than the traditional single-scale DenseNet, while the sample dropout method can further improve the classification robustness of multi-scale DenseNet.

Kele Xu, Dawei Feng, Haibo Mi et al.

Audio scene classification, the problem of predicting class labels of audio scenes, has drawn lots of attention during the last several years. However, it remains challenging and falls short of accuracy and efficiency. Recently, Convolutional Neural Network (CNN)-based methods have achieved better performance with comparison to the traditional methods. Nevertheless, conventional single channel CNN may fail to consider the fact that additional cues may be embedded in the multi-channel recordings. In this paper, we explore the use of Multi-channel CNN for the classification task, which aims to extract features from different channels in an end-to-end manner. We conduct the evaluation compared with the conventional CNN and traditional Gaussian Mixture Model-based methods. Moreover, to improve the classification accuracy further, this paper explores the using of mixup method. In brief, mixup trains the neural network on linear combinations of pairs of the representation of audio scene examples and their labels. By employing the mixup approach for data argumentation, the novel model can provide higher prediction accuracy and robustness in contrast with previous models, while the generalization error can also be reduced on the evaluation data.