Recent advancements in Large Language Models (LLMs) have shown that it is promising to utilize Process Reward Models (PRMs) as verifiers to enhance the performance of LLMs. However, current PRMs face three key challenges: (1) limited process supervision and generalization capabilities, (2) dependence on scalar value prediction without leveraging the generative abilities of LLMs, and (3) inability to scale the test-time compute of PRMs. In this work, we introduce GenPRM, a generative process reward model that performs explicit Chain-of-Thought (CoT) reasoning with code verification before providing judgment for each reasoning step. To obtain high-quality process supervision labels and rationale data, we propose Relative Progress Estimation (RPE) and a rationale synthesis framework that incorporates code verification. Experimental results on ProcessBench and several mathematical reasoning tasks show that GenPRM significantly outperforms prior PRMs with only 23K training data from MATH dataset. Through test-time scaling, a 1.5B GenPRM outperforms GPT-4o, and a 7B GenPRM surpasses Qwen2.5-Math-PRM-72B on ProcessBench. Additionally, GenPRM demonstrates strong abilities to serve as a critic model for policy model refinement. This work establishes a new paradigm for process supervision that bridges the gap between PRMs and critic models in LLMs.

Binary Neural Networks (BNNs) have garnered significant attention due to their immense potential for deployment on edge devices. However, the non-differentiability of the quantization function poses a challenge for the optimization of BNNs, as its derivative cannot be backpropagated. To address this issue, hypernetwork based methods, which utilize neural networks to learn the gradients of non-differentiable quantization functions, have emerged as a promising approach due to their adaptive learning capabilities to reduce estimation errors. However, existing hypernetwork based methods typically rely solely on current gradient information, neglecting the influence of historical gradients. This oversight can lead to accumulated gradient errors when calculating gradient momentum during optimization. To incorporate historical gradient information, we design a Historical Gradient Storage (HGS) module, which models the historical gradient sequence to generate the first-order momentum required for optimization. To further enhance gradient generation in hypernetworks, we propose a Fast and Slow Gradient Generation (FSG) method. Additionally, to produce more precise gradients, we introduce Layer Recognition Embeddings (LRE) into the hypernetwork, facilitating the generation of layer-specific fine gradients. Extensive comparative experiments on the CIFAR-10 and CIFAR-100 datasets demonstrate that our method achieves faster convergence and lower loss values, outperforming existing baselines.

As an effective approach to equip models with multitask capabilities without additional training, model merging has garnered significant attention. However, existing merging methods face challenges of redundant parameter conflicts and the excessive storage burden of fine-tuned parameters. In this work, through controlled experiments, we reveal that for fine-tuned task vectors, only those parameters with magnitudes above a certain threshold contribute positively to the task, exhibiting a pulse-like characteristic. We then attempt leveraging this pulse-like characteristic to binarize the task vectors and reduce storage overhead. Further controlled experiments show that the binarized task vectors incur almost no decrease in fine-tuning and merging performance, and even exhibit stronger performance improvements as the proportion of redundant parameters increases. Based on these insights, we propose Task Switch (T-Switch), which decomposes task vectors into three components: 1) an activation switch instantiated by a binarized mask vector, 2) a polarity switch instantiated by a binarized sign vector, and 3) a scaling knob instantiated by a scalar coefficient. By storing task vectors in a binarized form, T-Switch alleviates parameter conflicts while ensuring efficient task parameter storage. Furthermore, to enable automated switch combination in T-Switch, we further introduce Auto-Switch, which enables training-free switch combination via retrieval from a small query set. Experiments indicate that our methods achieve significant performance improvements over existing baselines, requiring only 1-3% of the storage space of full-precision parameters.

Black-Box Optimization (BBO) has found successful applications in many fields of science and engineering. Recently, there has been a growing interest in meta-learning particular components of BBO algorithms to speed up optimization and get rid of tedious hand-crafted heuristics. As an extension, learning the entire algorithm from data requires the least labor from experts and can provide the most flexibility. In this paper, we propose RIBBO, a method to reinforce-learn a BBO algorithm from offline data in an end-to-end fashion. RIBBO employs expressive sequence models to learn the optimization histories produced by multiple behavior algorithms and tasks, leveraging the in-context learning ability of large models to extract task information and make decisions accordingly. Central to our method is to augment the optimization histories with regret-to-go tokens, which are designed to represent the performance of an algorithm based on cumulative regret over the future part of the histories. The integration of regret-to-go tokens enables RIBBO to automatically generate sequences of query points that are positively correlated to the user-desired regret, verified by its universally good empirical performance on diverse problems, including BBO benchmark, hyper-parameter optimization, and robot control problems.

Bayesian optimization (BO) has been extensively studied and applied as a sample-efficient, black-box optimization method in neural network learning, particularly for hyperparameter optimization. However, high-dimensional Bayesian optimization (HDBO) remains a challenging research direction. While existing BO benchmarks focus on practical low-dimensional tasks and specific high-dimensional settings, current HDBO studies often rely on custom experimental settings, resulting in a lack of standardized benchmarks for comprehensive evaluation. To address this, we introduce the first standardized HDBO benchmark HDBO-B. HDBO-B encompasses diverse high-dimensional optimization tasks, synthetic functions tailored for common research settings, and a robust framework with standardized testing, method implementation, and extensive examples. By benchmarking nine representative optimization methods, we demonstrate HDBO-B’s validity and highlight new research directions in neural network methods. Codes are available at: https://github.com/Yiyuiii/HDBO-B.

Heterogeneous Large Language Model (LLM) fusion integrates the strengths of multiple source LLMs with different architectures into a target LLM with low computational overhead. While promising, existing methods suffer from two major limitations: 1) **reliance on real data from limited domain** for knowledge fusion, preventing the target LLM from fully acquiring knowledge across diverse domains, and 2) **fixed data allocation proportions** across domains, failing to dynamically adjust according to the target LLM's varying capabilities across domains, leading to a capability imbalance. To overcome these limitations, we propose Bohdi, a synthetic-data-only heterogeneous LLM fusion framework. Through the organization of knowledge domains into a hierarchical tree structure, Bohdi enables automatic domain exploration and multi-domain data generation through multi-model collaboration, thereby comprehensively extracting knowledge from source LLMs. By formalizing domain expansion and data sampling proportion allocation on the knowledge tree as a Hierarchical Multi-Armed Bandit problem, Bohdi leverages the designed DynaBranches mechanism to adaptively adjust sampling proportions based on the target LLM's performance feedback across domains. Integrated with our proposed Introspection-Rebirth (IR) mechanism, DynaBranches dynamically tracks capability shifts during target LLM's updates via Sliding Window Binomial Likelihood Ratio Testing (SWBLRT), further enhancing its online adaptation capability. Comparative experimental results on a comprehensive suite of benchmarks demonstrate that Bohdi significantly outperforms existing baselines on multiple target LLMs, exhibits higher data efficiency, and virtually eliminates the imbalance in the target LLM's capabilities.

Few-Shot Class-Incremental Learning (FSCIL) has gained considerable attention in recent years for its pivotal role in addressing continuously arriving classes. However, it encounters additional challenges. The scarcity of samples in new sessions intensifies overfitting, causing incompatibility between the output features of new and old classes, thereby escalating catastrophic forgetting. A prevalent strategy involves mitigating catastrophic forgetting through the Explicit Memory (EM), which comprise of class prototypes. However, current EM-based methods retrieves memory globally by performing Vector-to-Vector (V2V) interaction between features corresponding to the input and prototypes stored in EM, neglecting the geometric structure of local features. This hinders the accurate modeling of their positional relationships. To incorporate information of local geometric structure, we extend the V2V interaction to Graph-to-Graph (G2G) interaction. For enhancing local structures for better G2G alignment and the prevention of local feature collapse, we propose the Local Graph Preservation (LGP) mechanism. Additionally, to address sample scarcity in classes from new sessions, the Contrast-Augmented G2G (CAG2G) is introduced to promote the aggregation of same class features thus helps few-shot learning. Extensive comparisons on CIFAR100, CUB200, and the challenging ImageNet-R dataset demonstrate the superiority of our method over existing methods.

Variational Autoencoder based Bayesian Optimization (VAE-BO) has demonstrated its excellent performance in addressing high-dimensional structured optimization problems. However, current mainstream methods overlook the potential of utilizing a pool of unlabeled data to construct the latent space, while only concentrating on designing sophisticated models to leverage the labeled data. Despite their effective usage of labeled data, these methods often require extra network structures, additional procedure, resulting in computational inefficiency. To address this issue, we propose a novel method to effectively utilize unlabeled data with the guidance of labeled data. Specifically, we tailor the pseudo-labeling technique from semi-supervised learning to explicitly reveal the relative magnitudes of optimization objective values hidden within the unlabeled data. Based on this technique, we assign appropriate training weights to unlabeled data to enhance the construction of a discriminative latent space. Furthermore, we treat the VAE encoder and the Gaussian Process (GP) in Bayesian optimization as a unified deep kernel learning process, allowing the direct utilization of labeled data, which we term as Gaussian Process guidance. This directly and effectively integrates the goal of improving GP accuracy into the VAE training, thereby guiding the construction of the latent space. The extensive experiments demonstrate that our proposed method outperforms existing VAE-BO algorithms in various optimization scenarios. Our code will be published at https://github.com/TaicaiChen/PG-LBO.

AlphaZero and MuZero have achieved state-of-the-art (SOTA) performance in a wide range of domains, including board games and robotics, with discrete and continuous action spaces. However, to obtain an improved policy, they often require an excessively large number of simulations, especially for domains with large action spaces. As the simulation budget decreases, their performance drops significantly. In addition, many important real-world applications have combinatorial (or exponential) action spaces, making it infeasible to search directly over all possible actions. In this paper, we extend AlphaZero and MuZero to learn and plan in more complex multiagent (MA) Markov decision processes, where the action spaces increase exponentially with the number of agents. Our new algorithms, MA Gumbel AlphaZero and MA Gumbel MuZero, respectively without and with model learning, achieve superior performance on cooperative multiagent control problems, while reducing the number of environmental interactions by up to an order of magnitude compared to model-free approaches. In particular, we significantly improve prior performance when planning with much fewer simulation budgets. The code and appendix are available at https://github.com/tjuHaoXiaotian/MA-MuZero.

Attaining the equilibrium geometry of a catalyst-adsorbate system is key to fundamentally assessing its effective properties, such as adsorption energy. While machine learning methods with advanced representation or supervision strategies have been applied to boost and guide the relaxation processes of catalysis systems, existing methods that produce linearly aggregated geometry predictions are susceptible to edge representations ambiguity, and are therefore vulnerable to graph variations. In this paper, we present a novel graph neural network (GNN) supervision and prediction strategy DR-Label. Our approach mitigates the multiplicity of solutions in edge representation and encourages model predictions that are independent of graph structural variations. DR-Label first Deconstructs finer-grained equilibrium state information to the model by projecting the node-level supervision signal to each edge. Reversely, the model Reconstructs a more robust equilibrium state prediction by converting edge-level predictions to node-level via a sphere-fitting algorithm. When applied to three fundamentally different models, DR-Label consistently enhanced performance. Leveraging the graph structure invariance of the DR-Label strategy, we further propose DRFormer, which applied explicit intermediate positional update and achieves a new state-of-the-art performance on the Open Catalyst 2020 (OC20) dataset and the Cu-based single-atom alloys CO adsorption (SAA) dataset. We expect our work to highlight vital principles for advancing geometric GNN models for catalysis systems and beyond. Our code is available at https://github.com/bowenwang77/DR-Label

Learning representations to capture the very fundamental understanding of the world is a key challenge in machine learning. The hierarchical structure of explanatory factors hidden in data is such a general representation and could be potentially achieved with a hierarchical VAE. However, training a hierarchical VAE always suffers from the “posterior collapse”, where the data information is hard to propagate to the higher-level latent variables, hence resulting in a bad hierarchical representation. To address this issue, we first analyze the shortcomings of existing methods for mitigating the posterior collapse from an information theory perspective, then highlight the necessity of regularization for explicitly propagating data information to higher-level latent variables while maintaining the dependency between different levels. This naturally leads to formulating the inference of the hierarchical latent representation as a sequential decision process, which could benefit from applying reinforcement learning (RL). Aligning RL's objective with the regularization, we first introduce a skip-generative path to acquire a reward for evaluating the information content of an inferred latent representation, and then the developed Q-value function based on it could have a consistent optimization direction of the regularization. Finally, policy gradient, one of the typical RL methods, is employed to train a hierarchical VAE without introducing a gradient estimator. Experimental results firmly support our analysis and demonstrate that our proposed method effectively mitigates the posterior collapse issue, learns an informative hierarchy, acquires explainable latent representations, and significantly outperforms other hierarchical VAE-based methods in downstream tasks.

Advanced life forms, sustained by the synergistic interaction of neural cognitive mechanisms, continually acquire and transfer knowledge throughout their lifespan. In contrast, contemporary machine learning paradigms exhibit limitations in emulating the facets of continual learning (CL). Nonetheless, the emergence of large language models (LLMs) presents promising avenues for realizing CL via interactions with these models. Drawing on Complementary Learning System theory, this paper presents a novel Interactive Continual Learning (ICL) framework, enabled by collaborative interactions among models of various sizes. Specifically, we assign the ViT model as System1 and multimodal LLM as System2. To enable the memory module to deduce tasks from class information and enhance Set2Set retrieval, we propose the Class-Knowledge-Task Multi-Head Attention (CKT-MHA). Additionally, to improve memory retrieval in System1 through enhanced geometric representation, we introduce the CL-vMF mechanism, based on the von Mises-Fisher (vMF) distribution. Mean-while, we introduce the von Mises-Fisher Outlier Detection and Interaction (vMF-ODI) strategy to identify hard examples, thus enhancing collaboration between System1 and System2 for complex reasoning realization. Comprehensive evaluation of our proposed ICL demonstrates significant resistance to forgetting and superior performance relative to existing methods. Code is available at github.com/ICL.

Quality-Diversity (QD) algorithms, as a subset of evolutionary algorithms, have emerged as a powerful optimization paradigm with the aim of generating a set of high-quality and diverse solutions. Although QD has demonstrated competitive performance in reinforcement learning, its low sample efficiency remains a significant impediment for real-world applications. Recent research has primarily focused on augmenting sample efficiency by refining selection and variation operators of QD. However, one of the less considered yet crucial factors is the inherently large-scale issue of the QD optimization problem. In this paper, we propose a novel Cooperative Coevolution QD (CCQD) framework, which decomposes a policy network naturally into two types of layers, corresponding to representation and decision respectively, and thus simplifies the problem significantly. The resulting two (representation and decision) subpopulations are coevolved cooperatively. CCQD can be implemented with different selection and variation operators. Experiments on several popular tasks within the QDAX suite demonstrate that an instantiation of CCQD achieves approximately a 200% improvement in sample efficiency.

Model-based reinforcement learning (MBRL) holds the promise of sample-efficient learning by utilizing a world model, which models how the environment works and typically encompasses components for two tasks: observation modeling and reward modeling. In this paper, through a dedicated empirical investigation, we gain a deeper understanding of the role each task plays in world models and uncover the overlooked potential of sample-efficient MBRL by mitigating the domination of either observation or reward modeling. Our key insight is that while prevalent approaches of explicit MBRL attempt to restore abundant details of the environment via observation models, it is difficult due to the environment’s complexity and limited model capacity. On the other hand, reward models, while dominating implicit MBRL and adept at learning compact task-centric dynamics, are inadequate for sample-efficient learning without richer learning signals. Motivated by these insights and discoveries, we propose a simple yet effective approach, HarmonyDream, which automatically adjusts loss coefficients to maintain task harmonization, i.e. a dynamic equilibrium between the two tasks in world model learning. Our experiments show that the base MBRL method equipped with HarmonyDream gains 10%-69% absolute performance boosts on visual robotic tasks and sets a new state-of-the-art result on the Atari 100K benchmark. Code is available at https://github.com/thuml/HarmonyDream.

World models empower model-based agents to interactively explore, reason, and plan within imagined environments for real-world decision-making. However, the high demand for interactivity poses challenges in harnessing recent advancements in video generative models for developing world models at scale. This work introduces Interactive VideoGPT (iVideoGPT), a scalable autoregressive transformer framework that integrates multimodal signals—visual observations, actions, and rewards—into a sequence of tokens, facilitating an interactive experience of agents via next-token prediction. iVideoGPT features a novel compressive tokenization technique that efficiently discretizes high-dimensional visual observations. Leveraging its scalable architecture, we are able to pre-train iVideoGPT on millions of human and robotic manipulation trajectories, establishing a versatile foundation that is adaptable to serve as interactive world models for a wide range of downstream tasks. These include action-conditioned video prediction, visual planning, and model-based reinforcement learning, where iVideoGPT achieves competitive performance compared with state-of-the-art methods. Our work advances the development of interactive general world models, bridging the gap between generative video models and practical model-based reinforcement learning applications. Code and pre-trained models are available at https://thuml.github.io/iVideoGPT.

Knowledge transfer in cooperative multi-agent reinforcement learning (MARL) has drawn increasing attention in recent years. Unlike generalizing policies in single-agent tasks, it is more important to consider coordination knowledge than individual knowledge in multi-agent transfer learning. However, most of the existing methods only focus on knowledge transfer of the individual agent policy, which leads to coordination bias and finally affects the final performance in cooperative MARL. In this paper, we propose a level-adaptive MARL framework called “LA-QTransformer”, to realize the knowledge transfer on the coordination level via efficiently decomposing the agent coordination into multi-level coalition patterns for different agents. Compatible with centralized training with decentralized execution (CTDE) regime, LA-QTransformer utilizes the Level- Adaptive Transformer to generate suitable coalition patterns and then realizes the credit assignment for each agent. Besides, to deal with unexpected changes in the number of agents in the coordination transfer phase, we design a policy network called “Population invariant agent with Transformer (PIT)” to adapt dynamic observation and action space. We evaluate the LAQTransformer and PIT in the StarCraft II micro-management benchmark by comparing them with several state-of-the-art MARL baselines. The experimental results demonstrate the superiority of LA-QTransformer and PIT and verify the feasibility of coordination knowledge transfer.

The latent world model, which efficiently represents high-dimensional observations within a latent space, has shown promise in reinforcement learning-based policies for visual control tasks. Due to a lack of clear environmental context comprehension, its applicability in a variety of contexts with unknown dynamics is constrained. We propose a prototypical context- aware dynamics (ProtoCAD) model to address this issue. This model captures local dynamics using temporally consistent latent contexts and aids generalization in visual control tasks. By grouping prototypes over historical experiences, ProtoCAD collects useful contextual information that improves model-based reinforcement learning dynamics generalization in two ways. First, to guarantee the consistency of prototype assignments for various temporal segments of the same latent trajectory, a temporally consistent prototypes regularizer is used. Then, a context representation is devised to combine the aggregated prototype with the projection embedding of latent states. According to extensive trials, ProtoCAD outperforms competing approaches in terms of dynamics generalization for visual robotic control and autonomous driving applications.

Multi-types of behaviors (e.g., clicking, carting, purchasing, etc.) widely exist in most real-world recommendation scenarios, which are beneficial to learn users’ multi-faceted preferences. As dependencies are explicitly exhibited by the multiple types of behaviors, effectively modeling complex behavior dependencies is crucial for multi-behavior prediction. The state-of-the-art multi-behavior models learn behavior dependencies indistinguishably with all historical interactions as input. However, different behaviors may reflect different aspects of user preference, which means that some irrelevant interactions may play as noises to the target behavior to be predicted. To address the aforementioned limitations, we introduce multi-interest learning to the multi-behavior recommendation. More specifically, we propose a novel Coarse-to-fine Knowledge-enhanced Multi-interest Learning (CKML) framework to learn shared and behavior-specific interests for different behaviors. CKML introduces two advanced modules, namely Coarse-grained Interest Extracting (CIE) and Fine-grained Behavioral Correlation (FBC) , which work jointly to capture fine-grained behavioral dependencies. CIE uses knowledge-aware information to extract initial representations of each interest. FBC incorporates a dynamic routing scheme to further assign each behavior among interests. Empirical results on three real-world datasets verify the effectiveness and efficiency of our model in exploiting multi-behavior data.

Knowledge graphs (KGs) can help enhance recommendations, especially for the data-sparsity scenarios with limited user-item interaction data. Due to the strong power of representation learning of graph neural networks (GNNs), recent works of KG-based recommendation deploy GNN models to learn from both knowledge graph and user-item bipartite interaction graph. However, these works have not well considered the hierarchical structure of knowledge graph, leading to sub-optimal results. Despite the benefit of hierarchical structure, leveraging it is challenging since the structure is always partly-observed. In this work, we first propose to reveal unknown hierarchical structures with a supervised signal detection method and then exploit the hierarchical structure with disentangling representation learning. We conduct experiments on two large-scale datasets, of which the results well verify the superiority and rationality of the proposed method. Further experiments of ablation study with respect to key model designs have demonstrated the effectiveness and rationality of our proposed model. The code is available at https://github.com/tsinghua-fib-lab/HIKE .

The recommender system is one of the most fundamental information services. A significant effort has been devoted to improving prediction accuracy, inevitably leading to the potential degradation of recommendation diversity. Moreover, individuals have different needs for diversity. To address these problems, diversity-enhanced approaches are proposed to modify the recommender models. However, these methods fail to break free from the relevance-oriented paradigm and are mostly haunted by sharply-declined accuracy and high computational costs. To tackle these challenges, we propose the Dual-Process Graph Neural Network (DPGNN), an efficient diversity-enhanced recommender system, resonating with the dual-process model of human cognition and the arousal theory of human interest. The first stage reduces the risk of suboptimal output during the training procedure, which helps to find a solution outside the relevance-oriented paradigm. Moreover, the second stage utilizes user-specific rating adjustments, boosting the recommendation diversity and accommodating users' distinctive needs with minimum computational costs. Extensive experiments on real-world datasets verify the effectiveness of our method in improving diversity, while maintaining accuracy with low computational costs.

Multi-agent systems require effective coordination between groups and individuals to achieve common goals. However, current multi-agent reinforcement learning (MARL) methods primarily focus on improving individual policies and do not adequately address group-level policies, which leads to weak cooperation. To address this issue, we propose a novel Consensus-oriented Strategy (CoS) that emphasizes group and individual policies simultaneously. Specifically, CoS comprises two main components: (a) the vector quantized group consensus module, which extracts discrete latent embeddings that represent the stable and discriminative group consensus, and (b) the group consensus-oriented strategy, which integrates the group policy using a hypernet and the individual policies using the group consensus, thereby promoting coordination at both the group and individual levels. Through empirical experiments on cooperative navigation tasks with both discrete and continuous spaces, as well as google research football, we demonstrate that CoS outperforms state-of-the-art MARL algorithms and achieves better collaboration, thus providing a promising solution for achieving effective coordination in multi-agent systems.

Offline Reinforcement Learning (RL) aims to address the challenge of distribution shift between the dataset and the learned policy, where the value of out-of-distribution (OOD) data may be erroneously estimated due to overgeneralization. It has been observed that a considerable portion of the benefits derived from the conservative terms designed by existing offline RL approaches originates from their impact on the learned representation. This observation prompts us to scrutinize the learning dynamics of offline RL, formalize the process of generalization, and delve into the prevalent overgeneralization issue in offline RL. We then investigate the potential to rein the generalization from the representation perspective to enhance offline RL. Finally, we present Representation Distinction (RD), an innovative plug-in method for improving offline RL algorithm performance by explicitly differentiating between the representations of in-sample and OOD state-action pairs generated by the learning policy. Considering scenarios in which the learning policy mirrors the behavioral policy and similar samples may be erroneously distinguished, we suggest a dynamic adjustment mechanism for RD based on an OOD data generator to prevent data representation collapse and further enhance policy performance. We demonstrate the efficacy of our approach by applying RD to specially-designed backbone algorithms and widely-used offline RL algorithms. The proposed RD method significantly improves their performance across various continuous control tasks on D4RL datasets, surpassing several state-of-the-art offline RL algorithms.

Graph Neural Network (GNN)-based models have become the mainstream approach for recommender systems. Despite the effectiveness, they are still suffering from the cold-start problem, i.e., recommend for few-interaction items. Existing GNN-based recommendation models to address the cold-start problem mainly focus on utilizing auxiliary features of users and items, leaving the user-item interactions under-utilized. However, embeddings distributions of cold and warm items are still largely different, since cold items' embeddings are learned from lower-popularity interactions, while warm items' embeddings are from higher-popularity interactions. Thus, there is a seesaw phenomenon, where the recommendation performance for the cold and warm items cannot be improved simultaneously. To this end, we proposed a Uncertainty-aware Consistency learning framework for Cold-start item recommendation (shorten as UCC) solely based on user-item interactions. Under this framework, we train the teacher model (generator) and student model (recommender) with consistency learning, to ensure the cold items with additionally generated low-uncertainty interactions can have similar distribution with the warm items. Therefore, the proposed framework improves the recommendation of cold and warm items at the same time, without hurting any one of them. Extensive experiments on benchmark datasets demonstrate that our proposed method significantly outperforms state-of-the-art methods on both warm and cold items, with an average performance improvement of 27.6%.

In the information-overloaded era of the Web, recommender systems that provide personalized content filtering are now the mainstream portal for users to access Web information. Recommender systems deploy machine learning models to learn users’ preferences from collected historical data, leading to more centralized recommendation results due to the feedback loop. As a result, it will harm the ranking of content outside the narrowed scope and limit the options seen by users. In this work, we first conduct data analysis from a graph view to observe that the users’ feedback is restricted to limited items, verifying the phenomenon of centralized recommendation. We further develop a general simulation framework to derive the procedure of the recommender system, including data collection, model learning, and item exposure, which forms a loop. To address the filter bubble issue under the feedback loop, we then propose a general and easy-to-use reinforcement learning-based method, which can adaptively select few but effective connections between nodes from different communities as the exposure list. We conduct extensive experiments in the simulation framework based on large-scale real-world datasets. The results demonstrate that our proposed reinforcement learning-based control method can serve as an effective solution to alleviate the filter bubble and the separated communities induced by it. We believe the proposed framework of controllable recommendation in this work can inspire not only the researchers of recommender systems, but also a broader community concerned with artificial intelligence algorithms’ impact on humanity, especially for those vulnerable populations on the Web.

We study Policy-extended Value Function Approximator (PeVFA) in Reinforcement Learning (RL), which extends conventional value function approximator (VFA) to take as input not only the state (and action) but also an explicit policy representation. Such an extension enables PeVFA to preserve values of multiple policies at the same time and brings an appealing characteristic, i.e., value generalization among policies. We formally analyze the value generalization under Generalized Policy Iteration (GPI). From theoretical and empirical lens, we show that generalized value estimates offered by PeVFA may have lower initial approximation error to true values of successive policies, which is expected to improve consecutive value approximation during GPI. Based on above clues, we introduce a new form of GPI with PeVFA which leverages the value generalization along policy improvement path. Moreover, we propose a representation learning framework for RL policy, providing several approaches to learn effective policy embeddings from policy network parameters or state-action pairs. In our experiments, we evaluate the efficacy of value generalization offered by PeVFA and policy representation learning in several OpenAI Gym continuous control tasks. For a representative instance of algorithm implementation, Proximal Policy Optimization (PPO) re-implemented under the paradigm of GPI with PeVFA achieves about 40% performance improvement on its vanilla counterpart in most environments.

Recent algorithms designed for multi-agent tasks focus on finding a single optimal solution for all the agents. However, in many tasks (e.g., matrix games and transportation dispatching), there may exist more than one optimal solution, while previous algorithms can only converge to one of them. In many practical applications, it is important to develop reasonable agents with diverse behaviors. In this paper, we propose ”variational multi-agent policy diversification” (VMAPD), an on-policy framework for discovering diverse policies for coordination patterns of multiple agents. By taking advantage of latent variables and exploiting the connection between variational inference and multi-agent reinforcement learning, we derive a tractable evidence lower bound (ELBO) on the trajectories of all agents. Our algorithm uses policy iteration to maximize the derived lower bound and can be simply implemented by adding a pseudo reward during centralized learning. And the trained agents do not need to access the pseudo reward during decentralized execution. We demonstrate the effectiveness of our algorithm on several popular multi-agent testbeds. Experimental results show that VMAPD finds more solutions with similar sample complexity compared with other baselines.

Precise congestion prediction from a placement solution plays a crucial role in circuit placement. This work proposes the lattice hypergraph (LH-graph), a novel graph formulation for circuits, which preserves netlist data during the whole learning process, and enables the congestion information propagated geometrically and topologically. Based on the formulation, we further developed a heterogeneous graph neural network architecture LHNN, jointing the routing demand regression to support the congestion spot classification. LHNN constantly achieves more than 35% improvements compared with U-nets and Pix2Pix on the F1 score. We expect our work shall highlight essential procedures using machine learning for congestion prediction.

Graph neural networks (GNNs) have achieved significant success in numerous fields under settings where training and testing graphs are identically distributed. However, this setting is rarely satisfied in real life. Due to the lack of out-of-distribution (OOD) generalization abilities, existing GNNs methods perform disappointingly when there exist distribution shifts between testing and training graphs. Though several attempts have been made to deal with the issue, they mainly focus on structural properties while overlooking rich graph feature information. To this end, we propose an Invariant Factor GNN (IFGNN), which utilizes causal factor graphs to achieve invariant performances across different environments. Specifically, we dissect the graph generalization problem in a causal view, and argue that the key of graph generalization lies in discovering causal factors. Thus we extract the latent factors in the graph through disentanglement, and the causal ones are discovered with the invariant learning mechanism. We conduct extensive experiments on both synthetic and real-world datasets with distribution shifts to validate the OOD generalization abilities. The results demonstrate that our proposed IFGNN significantly outperforms the state-of-the-art baselines.

We propose Neuro-Symbolic Hierarchical Rule Induction, an efficient interpretable neuro-symbolic model, to solve Inductive Logic Programming (ILP) problems. In this model, which is built from a pre-defined set of meta-rules organized in a hierarchical structure, first-order rules are invented by learning embeddings to match facts and body predicates of a meta-rule. To instantiate, we specifically design an expressive set of generic meta-rules, and demonstrate they generate a consequent fragment of Horn clauses. As a differentiable model, HRI can be trained both via supervised learning and reinforcement learning. To converge to interpretable rules, we inject a controlled noise to avoid local optima and employ an interpretability-regularization term. We empirically validate our model on various tasks (ILP, visual genome, reinforcement learning) against relevant state-of-the-art methods, including traditional ILP methods and neuro-symbolic models.

Offline reinforcement learning is a promising approach for practical applications since it does not require interactions with real-world environments. However, existing offline RL methods only work well in environments with continuous or small discrete action spaces. In environments with large and discrete action spaces, such as recommender systems and dialogue systems, the performance of existing methods decreases drastically because they suffer from inaccurate value estimation for a large proportion of out-of-distribution (o.o.d.) actions. While recent works have demonstrated that online RL benefits from incorporating semantic information in action representations, unfortunately, they fail to learn reasonable relative distances between action representations, which is key to offline RL to reduce the influence of o.o.d. actions. This paper proposes an action representation learning framework for offline RL based on a pseudometric, which measures both the behavioral relation and the data-distributional relation between actions. We provide theoretical analysis on the continuity of the expected Q-values and the offline policy improvement using the learned action representations. Experimental results show that our methods significantly improve the performance of two typical offline RL methods in environments with large and discrete action spaces.

Due to the rapid growth in the scale of circuits and the desire for knowledge transfer from old designs to new ones, deep learning technologies have been widely exploited in Electronic Design Automation (EDA) to assist circuit design. In chip design cycles, we might encounter heterogeneous and diverse information sources, including the two most informative ones: the netlist and the design layout. However, handling each information source independently is sub-optimal. In this paper, we propose a novel way to integrate the multiple information sources under a unified heterogeneous graph named Circuit Graph, where topological and geometrical information is well integrated. Then, we propose Circuit GNN to fully utilize the features of vertices, edges as well as heterogeneous information during the message passing process. It is the first attempt to design a versatile circuit representation that is compatible across multiple EDA tasks and stages. Experiments on the two most representative prediction tasks in EDA show that our solution reaches state-of-the-art performance in both logic synthesis and global placement chip design stages. Besides, it achieves a 10x speed-up on congestion prediction compared to the state-of-the-art model.

Context, the embedding of previous collected trajectories, is a powerful construct for Meta-Reinforcement Learning (Meta-RL) algorithms. By conditioning on an effective context, Meta-RL policies can easily generalize to new tasks within a few adaptation steps. We argue that improving the quality of context involves answering two questions: 1. How to train a compact and sufficient encoder that can embed the task-specific information contained in prior trajectories? 2. How to collect informative trajectories of which the corresponding context reflects the specification of tasks? To this end, we propose a novel Meta-RL framework called CCM (Contrastive learning augmented Context-based Meta-RL). We first focus on the contrastive nature behind different tasks and leverage it to train a compact and sufficient context encoder. Further, we train a separate exploration policy and theoretically derive a new information-gain-based objective which aims to collect informative trajectories in a few steps. Empirically, we evaluate our approaches on common benchmarks as well as several complex sparse-reward environments. The experimental results show that CCM outperforms state-of-the-art algorithms by addressing previously mentioned problems respectively.

Meta-Reinforcement Learning (meta-RL) algorithms enable agents to adapt to new tasks from small amounts of exploration, based on the experience of similar tasks. Recent studies have pointed out that a good representation of a task is key to the success of off-policy context-based meta-RL. Inspired by contrastive methods in unsupervised representation learning, we propose a new method to learn the task representation based on the mutual information between transition tuples in a trajectory and the task embedding. We also propose a new estimation for task similarity based on Q-function, which can be used to form a constraint on the distribution of the encoded task variables, making the task encoder encode the task variables more effective on new tasks. Experiments on meta-RL tasks show that the newly proposed method outperforms existing meta-RL algorithms.

Practical recommender systems experience a cold-start problem when observed user-item interactions in the history are insufficient. Meta learning, especially gradient based one, can be adopted to tackle this problem by learning initial parameters of the model and thus allowing fast adaptation to a specific task from limited data examples. Though with significant performance improvement, it commonly suffers from two critical issues: the non-compatibility with mainstream industrial deployment and the heavy computational burdens, both due to the inner-loop gradient operation. These two issues make them hard to be applied in practical recommender systems. To enjoy the benefits of meta learning framework and mitigate these problems, we propose a recommendation framework called Contextual Modulation Meta Learning (CMML). CMML is composed of fully feed-forward operations so it is computationally efficient and completely compatible with the mainstream industrial deployment. CMML consists of three components, including a context encoder that can generate context embedding to represent a specific task, a hybrid context generator that aggregates specific user-item features with task-level context, and a contextual modulation network, which can modulate the recommendation model to adapt effectively. We validate our approach on both scenario-specific and user-specific cold-start setting on various real-world datasets, showing CMML can achieve comparable or even better performance with gradient based methods yet with higher computational efficiency and better interpretability.

Presently with technology node scaling, an accurate prediction model at early design stages can significantly reduce the design cycle. Especially during logic synthesis, predicting cell congestion due to improper logic combination can reduce the burden of subsequent physical implementations. There have been attempts using Graph Neural Network (GNN) techniques to tackle congestion prediction during the logic synthesis stage. However, they require informative cell features to achieve reasonable performance since the core idea of GNNs is built on the message passing framework, which would be impractical at the early logic synthesis stage. To address this limitation, we propose a framework that can directly learn embeddings for the given netlist to enhance the quality of our node features. Popular random-walk based embedding methods such as Node2vec, LINE, and DeepWalk suffer from the issue of cross-graph alignment and poor generalization to unseen netlist graphs, yielding inferior performance and costing significant runtime. In our framework, we introduce a superior alternative to obtain node embeddings that can generalize across netlist graphs using matrix factorization methods. We propose an efficient mini-batch training method at the sub-graph level that can guarantee parallel training and satisfy the memory restriction for large-scale netlists. We present results utilizing open-source EDA tools such as DREAMPLACE and OPENROAD frameworks on a variety of openly available circuits. By combining the learned embedding on top of the netlist with the GNNs, our method improves prediction performance, generalizes to new circuit lines, and is efficient in training, potentially saving over 90% of runtime.

The partially observable Markov decision process (POMDP) provides a general framework for modeling an agent's decision process with state uncertainty, and online planning plays a pivotal role in solving it. A belief is a distribution of states representing state uncertainty. Methods for large-scale POMDP problems rely on the same idea of sampling both states and observations. That is, instead of exact belief updating, a collection of sampled states is used to approximate the belief; instead of considering all possible observations, only a set of sampled observations are considered. Inspired by this, we take one step further and propose an online planning algorithm, Adaptive Online Packing-guided Search (AdaOPS), to better approximate beliefs with adaptive particle filter technique and balance estimation bias and variance by fusing similar observation branches. Theoretically, our algorithm is guaranteed to find an $\epsilon$-optimal policy with a high probability given enough planning time under some mild assumptions. We evaluate our algorithm on several tricky POMDP domains, and it outperforms the state-of-the-art in all of them.

Transfer Learning has shown great potential to enhance single-agent Reinforcement Learning (RL) efficiency. Similarly, Multiagent RL (MARL) can also be accelerated if agents can share knowledge with each other. However, it remains a problem of how an agent should learn from other agents. In this paper, we propose a novel Multiagent Policy Transfer Framework (MAPTF) to improve MARL efficiency. MAPTF learns which agent's policy is the best to reuse for each agent and when to terminate it by modeling multiagent policy transfer as the option learning problem. Furthermore, in practice, the option module can only collect all agent's local experiences for update due to the partial observability of the environment. While in this setting, each agent's experience may be inconsistent with each other, which may cause the inaccuracy and oscillation of the option-value's estimation. Therefore, we propose a novel option learning algorithm, the successor representation option learning to solve it by decoupling the environment dynamics from rewards and learning the option-value under each agent's preference. MAPTF can be easily combined with existing deep RL and MARL approaches, and experimental results show it significantly boosts the performance of existing methods in both discrete and continuous state spaces.

Social psychology and real experiences show that cognitive consistency plays an important role to keep human society in order: if people have a more consistent cognition about their environments, they are more likely to achieve better cooperation. Meanwhile, only cognitive consistency within a neighborhood matters because humans only interact directly with their neighbors. Inspired by these observations, we take the first step to introduce neighborhood cognitive consistency (NCC) into multi-agent reinforcement learning (MARL). Our NCC design is quite general and can be easily combined with existing MARL methods. As examples, we propose neighborhood cognition consistent deep Q-learning and Actor-Critic to facilitate large-scale multi-agent cooperations. Extensive experiments on several challenging tasks (i.e., packet routing, wifi configuration and Google football player control) justify the superior performance of our methods compared with state-of-the-art MARL approaches.

This paper proposes a novel vision-based object detection method in autonomous driving, which introduces the temporal information into the deep learning-based detection method for moving object detection. Vision-based object detection is a critical technology for autonomous driving. The objects in the real world such as driving cars, don't have great changes in their positions and velocities. So the position change of objects between two consecutive frames is not large. This is usually ignored by traditional works, which usually use object detection methods on still-images to detect moving objects. Considering the relationship among consecutive frames (temporal information), we present a robust and real-time tracking method following image detection to refine the object detection results. Based on the three key attributes (distances, sizes and positions), the tracking method aims to build the association between the detected objects on the current frame and those in previous frames. The proposed object detection with temporal information dramatically improves the performance of existing object detection algorithms based on stillimage. With the proposed method, we won the champion in the preceding vehicle detection task in 2017 intelligent vehicle future challenge(2017 IVFC)1.

This paper investigates the traffic sign recognition task with deep learning methods. The proposed algorithm which is called DeepSign includes three modules: a detection module (PosNet) for locating the traffic sign in a static image, a classification module (PatchNet) for classifying the detected image patch, and a temporal filter for correcting the recognition results. The PosNet is a binary object detection convolution neural network which regards all traffic signs as one class and the background as the other class. Different from the traditional works which recognize the traffic sign on the static image, the proposed temporal filter exploits the contextual information to recover the missed detection region and correct the false classification. The experiments validate the effectiveness of the proposed algorithm. It achieved the third place on the traffic sign recognition task in 2017 China intelligent vehicle future challenge (2017 CIVFC).

Policy gradient algorithm is often used to deal with the continuous control problems. But as a model-free algorithm, it suffers from the low data efficiency and long learning phase. In this paper, a policy gradient with Gaussian process modelling (PGGPM) algorithm is proposed to accelerate learning process. The system model is approximated by Gaussian process in an incremental way, which is used to explore state action space virtually by generating imaginary samples. Both the real and imaginary samples are used to train the actor and critic networks. Finally, we apply our algorithm to two experiments to verify that Gaussian process can accurately fit system model and the supplementary imaginary samples can speed up the learning phase.

This paper combines an efficient reinforcement learning algorithm named Multisamples in Each Cell (MEC) with a building thermal comfort control problem. It implements the efficient exploration rule and makes high use of observed samples. A grid is utilized to partition the continuous state into cells that are used to store samples. A near-upper Q function is obtained based on the samples in each cell. The value iteration technique is designed to derive the near optimal control policy. The algorithm can efficiently balance exploration and exploitation. The entire implementation process needs no model of systems. The thermal comfort criterion, predicted mean vote, is introduced to evaluate zone thermal comfort status. A two story, multi-zone small office building equipped with a variable air volume direct expansion cooling system is built in EnergyPlus to establish an EnergyPlus-MATLAB co-simulation platform. A MEC thermal comfort control simulation is implemented to validate the high performance property compared with Q-learning.