Large language models (LLMs) have shown promise in providing scalable mental health support, while evaluating their counseling capability remains crucial to ensure both efficacy and safety. Existing evaluations are limited by the static assessment that focuses on knowledge tests, the single perspective that centers on user experience, and the open-loop framework that lacks actionable feedback. To address these issues, we propose Ψ-Arena, an interactive framework for comprehensive assessment and optimization of LLM-based counselors, featuring three key characteristics: (1) Realistic arena interactions that simulate real-world counseling through multi-stage dialogues with psychologically profiled NPC clients; (2) Tripartite evaluation that integrates assessments from the client, supervisor, and counselor perspectives; (3) Closed-loop optimization that iteratively improves LLM counselors using diagnostic feedback. Experiments across eight state-of-the-art LLMs show significant performance variations in different real-world scenarios and evaluation perspectives. Moreover, reflection-based optimization results in up to a 141% improvement in counseling performance. We hope Ψ-Arena provides a foundational resource for advancing reliable and human-aligned LLM applications in mental healthcare.

The video game industry has emerged as a significant economic force, driving extensive research on optimizing the gaming environment and improving gaming experiences. Among these endeavors, player behavior representation learning has become a critical way to model valuable player properties and is beneficial for a wide range of downstream tasks. However, some common factors, such as login rewards and daily tasks, can trigger similar behaviors among different players, which are informative and noisy for learning high-quality player behavior representations. Existing methods ignore the low signal-to-noise ratio in player behavior data and waste too much modeling capacity on less informative behaviors, resulting in their learned representations being noisy. In this paper, we propose a novel model for Denoised Player Behavior representation Learning, namely DPBL, which consists of two key modules. The first module extracts various player behavior patterns and isolates them from less informative noise. The second module utilizes the extracted patterns to refine the embedding of each behavior and eliminates noise. To optimize DPBL, two contrastive learning strategies are proposed to identify the noise that should be eliminated and to learn distinguishable representations, respectively. With the above design, DPBL is capable of mitigating the impact of noise in the data and learning high-quality representations that effectively capture player characteristics. We conducted extensive experiments on two real-world datasets, and DPBL outperforms all baselines on various downstream tasks with an improvement of 1.4% ∼ 18.1%. The results also show that DPBL achieves an improvement of 5.6% ∼ 23.0% in the denoising experiments, which proves that DPBL is more robust to noisy behaviors. Code is available at https://github.com/LwbXc/DPBL.

Story visualization has gained increasing attention in artificial intelligence. However, existing methods still struggle with maintaining a balance between character identity preservation and text-semantics alignment, largely due to a lack of detailed semantic modeling of the story scene. To tackle this challenge, we propose a novel knowledge graph, namely Character-Graph (CG), which represents various story-related knowledge, including the characters, their attributes and the relationship. We then introduce StoryWeaver, an image generator that achieves Customization via Character-Graph (C-CG), capable of consistent story visualization with rich text semantics. To further improve the multi-character generation performance, we incorporate knowledge-enhanced spatial guidance (KE-SG) into StoryWeaver to precisely inject character semantics into generation. To validate the effectiveness of our proposed method, extensive experiments are conducted using a new benchmark called TBC-Bench. The experiments confirm that our StoryWeaver excels not only in creating vivid visual story plots but also in accurately conveying character identities across various scenarios with considerable storage efficiency, e.g., achieving an average increase of +9.03% DINO-I and +13.44% CLIP-T. Furthermore, ablation experiments are conducted to verify the superiority of each proposed module.

Character-based dialogue (aka role-playing) enables users to freely customize characters for interaction, which often relies on LLMs, raising the need to evaluate LLMs’ character customization capability. However, existing benchmarks fail to ensure a robust evaluation as they often only involve a single character category or evaluate limited dimensions. Moreover, the sparsity of character features in responses makes feature-focused generative evaluation both ineffective and inefficient. To address these issues, we propose CharacterBench, the largest bilingual generative benchmark, with 22,859 human-annotated samples covering 3,956 characters from 25 detailed character categories. We define 11 dimensions of 6 aspects, classified as sparse and dense dimensions based on whether character features evaluated by specific dimensions manifest in each response. We enable effective and efficient evaluation by crafting tailored queries for each dimension to induce characters’ responses related to specific dimensions. Further, we develop CharacterJudge model for cost-effective and stable evaluations. Experiments show its superiority over SOTA automatic judges (e.g., GPT-4) and our benchmark’s potential to optimize LLMs’ character customization.

Personality is an important concept in psychology that reflects individual differences in thinking and behavior, and has significant applications across various fields.Most existing personality analysis methods address this issue at the bag level, treating the entire corpus gathered from one individual as a single unit for classification.However, this paradigm presents several challenges.From the data perspective, collecting a large corpus for each individual and performing comprehensive annotations pose significant difficulties in both data collection and labeling.On the application side, concentrating on classifying the entire corpus limits its applicability in more common single-instance scenarios.To address these issues, we propose a new task paradigm in text-based personality representation learning.Specifically, we construct a triplet personality trend comparison dataset to learn single-sentence personality embeddings with desirable metric properties.This approach removes the traditional constraints on data sources, facilitating dataset expansion, and can leverage the transfer capabilities of embeddings to easily adapt to various downstream tasks.Our experiments show that the learned embeddings significantly boost performance by a relative 10% across various applications, including personality detection, personality retrieval, and emotion translation prediction.The code and dataset are available at https://github.com/zjutangk/PTCD.

Jinfeng Zhou, Yuxuan Chen, Jianing Yin, Yongkang Huang, Yihan Shi, Xikun Zhang, Libiao Peng, Rongsheng Zhang, Tangjie Lv, Zhipeng Hu, Hongning Wang, Minlie Huang. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

In practice, reinforcement learning (RL) agents are often trained with a possibly imperfect proxy reward function, which may lead to a human-agent alignment issue (i.e., the learned policy either converges to non-optimal performance with low cumulative rewards, or achieves high cumulative rewards but in an undesired manner). To tackle this issue, we consider a framework where a human labeler can provide additional feedback in the form of corrective actions, which expresses the labeler's action preferences although this feedback may possibly be imperfect as well. In this setting, to obtain a better-aligned policy guided by both learning signals, we propose a novel value-based deep RL algorithm called **I**terative learning from **Co**rrective actions and **Pro**xy rewards (ICoPro), which cycles through three phases: (1) Solicit sparse corrective actions from a human labeler on the agent's demonstrated trajectories; (2) Incorporate these corrective actions into the Q-function using a margin loss to enforce adherence to labeler's preferences; (3) Train the agent with standard RL losses regularized with a margin loss to learn from proxy rewards and propagate the Q-values learned from human feedback. Moreover, another novel design in our approach is to integrate pseudo-labels from the target Q-network to reduce human labor and further stabilize training. We experimentally validate our proposition on a variety of tasks (Atari games and autonomous driving on highway). On the one hand, using proxy rewards with different levels of imperfection, our method can better align with human and is more sample-efficient than baseline methods. On the other hand, facing corrective actions with different types of imperfection, our method can overcome the non-optimality of this feedback thanks to the guidance from proxy rewards.

Within the domain of Massively Multiplayer Online (MMO) economy research, Agent-Based Modeling (ABM) has emerged as a robust tool for analyzing game economics, evolving from rule-based agents to decision-making agents enhanced by reinforcement learning. Nevertheless, existing works encounter significant challenges when attempting to emulate human-like economic activities among agents, particularly regarding agent reliability, sociability, and interpretability.In this study, we take a preliminary step in introducing a novel approach using Large Language Models (LLMs) in MMO economy simulation. Leveraging LLMs' role-playing proficiency, generative capacity, and reasoning aptitude, we design LLM-driven agents with human-like decision-making and adaptability. These agents are equipped with the abilities of role-playing, perception, memory, and reasoning, addressing the aforementioned challenges effectively. Simulation experiments focusing on in-game economic activities demonstrate that LLM-empowered agents can promote emergent phenomena like role specialization and price fluctuations in line with market rules.

Audio-driven talking head generation has drawn growing attention. To produce talking head videos with desired facial expressions, previous methods rely on extra reference videos to provide expression information, which may be difficult to find and hence limits their usage. In this work, we propose TalkCLIP, a framework that can generate talking heads where the expressions are specified by natural language, hence allowing for specifying expressions more conveniently. To model the mapping from text to expressions, we first construct a text-video paired talking head dataset where each video has diverse text descriptions that depict both coarse-grained emotions and fine-grained facial movements. Leveraging the proposed dataset, we introduce a CLIP-based style encoder that projects natural language-based descriptions to the representations of expressions. TalkCLIP can even infer expressions for descriptions unseen during training. TalkCLIP can also use text to modulate expression intensity and edit expressions. Extensive experiments demonstrate that TalkCLIP achieves the advanced capability of generating photo-realistic talking heads with vivid facial expressions guided by text descriptions.

Large-scale vision-language pre-training has achieved significant performance in multi-modal understanding and generation tasks. However, existing methods often perform poorly on image-text matching tasks that require structured representations, i.e., representations of objects, attributes, and relations. The models cannot make a distinction between "An astronaut rides a horse" and "A horse rides an astronaut". This is because they fail to fully leverage structured knowledge when learning multi-modal representations. In this paper, we present an end-to-end framework Structure-CLIP, which integrates Scene Graph Knowledge (SGK) to enhance multi-modal structured representations. Firstly, we use scene graphs to guide the construction of semantic negative examples, which results in an increased emphasis on learning structured representations. Moreover, a Knowledge-Enhance Encoder (KEE) is proposed to leverage SGK as input to further enhance structured representations. To verify the effectiveness of the proposed framework, we pre-train our model with the aforementioned approaches and conduct experiments on downstream tasks. Experimental results demonstrate that Structure-CLIP achieves state-of-the-art (SOTA) performance on VG-Attribution and VG-Relation datasets, with 12.5% and 4.1% ahead of the multi-modal SOTA model respectively. Meanwhile, the results on MSCOCO indicate that Structure-CLIP significantly enhances the structured representations while maintaining the ability of general representations. Our code is available at https://github.com/zjukg/Structure-CLIP.

In cooperative multi-agent reinforcement learning, decentralized agents hold the promise of overcoming the combinatorial explosion of joint action space and enabling greater scalability. However, they are susceptible to a game-theoretic pathology called relative overgeneralization that shadows the optimal joint action. Although recent value-decomposition algorithms guide decentralized agents by learning a factored global action value function, the representational limitation and the inaccurate sampling of optimal joint actions during the learning process make this problem still. To address this limitation, this paper proposes a novel algorithm called Optimistic Value Instructors (OVI). The main idea behind OVI is to introduce multiple optimistic instructors into the value-decomposition paradigm, which are capable of suggesting potentially optimal joint actions and rectifying the factored global action value function to recover these optimal actions. Specifically, the instructors maintain optimistic value estimations of per-agent local actions and thus eliminate the negative effects caused by other agents' exploratory or sub-optimal non-cooperation, enabling accurate identification and suggestion of optimal joint actions. Based on the instructors' suggestions, the paper further presents two instructive constraints to rectify the factored global action value function to recover these optimal joint actions, thus overcoming the RO problem. Experimental evaluation of OVI on various cooperative multi-agent tasks demonstrates its superior performance against multiple baselines, highlighting its effectiveness.

Matchmaking is a core task in e-sports and online games, as it contributes to player engagement and further influences the game's lifecycle. Previous methods focus on creating fair games at all times. They divide players into different tiers based on skill levels and only select players from the same tier for each game. Though this strategy can ensure fair matchmaking, it is not always good for player engagement. In this paper, we propose a novel Engagement-oriented Matchmaking (EnMatch) framework to ensure fair games and simultaneously enhance player engagement. Two main issues need to be addressed. First, it is unclear how to measure the impact of different team compositions and confrontations on player engagement during the game considering the variety of player characteristics. Second, such a detailed consideration on every single player during matchmaking will result in an NP-hard combinatorial optimization problem with non-linear objectives. In light of these challenges, we turn to real-world data analysis to reveal engagement-related factors. The resulting insights guide the development of engagement modeling, enabling the estimation of quantified engagement before a match is completed. To handle the combinatorial optimization problem, we formulate the problem into a reinforcement learning framework, in which a neural combinatorial optimization problem is built and solved. The performance of EnMatch is finally demonstrated through the comparison with other state-of-the-art methods based on several real-world datasets and online deployments on two games.

In human-centric content generation, the pre-trained text-to-image models struggle to produce user-wanted por-trait images, which retain the identity of individuals while exhibiting diverse expressions. This paper introduces our efforts towards personalized face generation. To this end, we propose a novel multi-modal face generation frame-work, capable of simultaneous identity-expression control and more fine-grained expression synthesis. Our expression control is so sophisticated that it can be specialized by the fine-grained emotional vocabulary. We devise a novel dif-fusion model that can undertake the task of simultaneously face swapping and reenactment. Due to the entanglement of identity and expression, separately and precisely control-ling them within one framework is a nontrivial task, thus has not been explored yet. To overcome this, we propose sev-eral innovative designs in the conditional diffusion model, including balancing identity and expression encoder, improved midpoint sampling, and explicitly background con-ditioning. Extensive experiments have demonstrated the controllability and scalability of the proposed framework, in comparison with state-of-the-art text-to-image, face swap-ping, and face reenactment methods.

Aligning agent behaviors with diverse human preferences remains a challenging problem in reinforcement learning (RL), owing to the inherent abstractness and mutability of human preferences. To address these issues, we propose AlignDiff, a novel framework that leverages RLHF to quantify human preferences, covering abstractness, and utilizes them to guide diffusion planning for zero-shot behavior customizing, covering mutability. AlignDiff can accurately match user-customized behaviors and efficiently switch from one to another. To build the framework, we first establish the multi-perspective human feedback datasets, which contain comparisons for the attributes of diverse behaviors, and then train an attribute strength model to predict quantified relative strengths. After relabeling behavioral datasets with relative strengths, we proceed to train an attribute-conditioned diffusion model, which serves as a planner with the attribute strength model as a director for preference aligning at the inference phase. We evaluate AlignDiff on various locomotion tasks and demonstrate its superior performance on preference matching, switching, and covering compared to other baselines. Its capability of completing unseen downstream tasks under human instructions also showcases the promising potential for human-AI collaboration. More visualization videos are released on https://aligndiff.github.io/.

Previous literature on policy diversity in reinforcement learning (RL) either focuses on the online setting or ignores the policy performance. In contrast, offline RL, which aims to learn high-quality policies from batched data, has yet to fully leverage the intrinsic diversity of the offline dataset. Addressing this dichotomy and aiming to balance quality and diversity poses a significant challenge to extant methodologies. This paper introduces a novel approach, termed Stylized Offline RL (SORL), which is designed to extract high-performing, stylistically diverse policies from a dataset characterized by distinct behavioral patterns. Drawing inspiration from the venerable Expectation-Maximization (EM) algorithm, SORL innovatively alternates between policy learning and trajectory clustering, a mechanism that promotes policy diversification. To further augment policy performance, we introduce advantage-weighted style learning into the SORL framework. Experimental evaluations across multiple environments demonstrate the significant superiority of SORL over previous methods in extracting high-quality policies with diverse behaviors. A case in point is that SORL successfully learns strong policies with markedly distinct playing patterns from a real-world human dataset of a popular basketball video game "Dunk City Dynasty."

Offline reinforcement learning (RL) is crucial for real-world applications where exploration can be costly or unsafe. However, offline learned policies are often suboptimal, and further online fine-tuning is required. In this paper, we tackle the fundamental dilemma of offline-to-online fine-tuning: if the agent remains pessimistic, it may fail to learn a better policy, while if it becomes optimistic directly, performance may suffer from a sudden drop. We show that Bayesian design principles are crucial in solving such a dilemma. Instead of adopting optimistic or pessimistic policies, the agent should act in a way that matches its belief in optimal policies. Such a probability-matching agent can avoid a sudden performance drop while still being guaranteed to find the optimal policy. Based on our theoretical findings, we introduce a novel algorithm that outperforms existing methods on various benchmarks, demonstrating the efficacy of our approach. Overall, the proposed approach provides a new perspective on offline-to-online RL that has the potential to enable more effective learning from offline data.

As a core problem of online games, matchmaking is to assign players into multiple teams to maximize their gaming experience. With the rapid development of game industry, it is increasingly difficulty to explicitly model players' experiences as linear functions. Instead, it is often modeled in a data-driven way by training a neural network. Meanwhile, complex rules must be satisfied to ensure the robustness of matchmaking, which are often described using logical operators. Therefore, matchmaking in practical scenarios is a challenging combinatorial optimization problem with nonlinear objective, linear constraints and logical constraints, which receives much less attention in previous research. In this paper, we propose a novel deep learning method for high-quality matchmaking in real-time. We first cast the problem as standard mixed-integer programming (MIP) by linearizing ReLU networks and logical constraints. Then, based on supervised learning, we design and train a multi-modal graph learning architecture to predict optimal solutions end-to-end from instance data, and solve a surrogate problem to efficiently obtain feasible solutions. Evaluation results on real industry datasets show that our method can deliver near-optimal solutions within 100ms.

In recent years, uplift modeling, also known as individual treatment effect (ITE) estimation, has seen wide applications in online marketing, such as delivering one-time issuance of coupons or discounts to motivate users' purchases. However, complex yet more realistic scenarios involving multiple interventions over time on users are still rarely explored. The challenges include handling the bias from time-varying confounders, determining optimal treatment timing, and selecting among numerous treatments. In this paper, to tackle the aforementioned challenges, we present a temporal point process-based uplift model (TPPUM) that utilizes users' temporal event sequences to estimate treatment effects via counterfactual analysis and temporal point processes. In this model, marketing actions are considered as treatments, user purchases as outcome events, and how treatments alter the future conditional intensity function of generating outcome events as the uplift. Empirical evaluations demonstrate that our method outperforms existing baselines on both real-world and synthetic datasets. In the online experiment conducted in a discounted bundle recommendation scenario involving an average of 3 to 4 interventions per day and hundreds of treatment candidates, we demonstrate how our model outperforms current state-of-the-art methods in selecting the appropriate treatment and timing of treatment, resulting in a 3.6% increase in application-level revenue.

Intermittent control problems are common in real world. The interactions between the decision maker and the executor can be discontinuous (intermittent) due to various types of interruptions, e.g. unstable communication channel. Due to intermittent interaction, agents are unable to acquire the state sent by the executor and cannot transmit actions to the executor within a period of time step, i.e. bidirectional blockage, which may lead to inefficiencies of reinforcement learning policies and prevent the executors from completing the task. Such problem is not well studied in the RL community. In this paper, we model Intermittent control problem as an Intermittent Control Markov Decision Process, i.e agents are expected to generate action sequences corresponding to the unavailable states and transmit them before disabling interactions to ensure the smooth and effective motion of executors. However, directly generating multiple future actions in the original action space has unnatural motion issue and exploration difficulty. We propose **M**ulti-step **A**ction **R**epre**S**entation (**MARS**), which encodes a sequence of actions from the original action space to a compact and decodable latent space. Then based on the latent action sequence representation, the mainstream RL methods can be easily optimized to learn a smooth and efficient motion policy. Extensive experiments on simulation tasks and real-world robotic grasping tasks show that MARS significantly improves the learning efficiency and final performances compared with existing baselines.

Hybrid action control tasks are common in the real world, which require controlling some discrete and continuous actions simultaneously. To solve these tasks, existing Deep Reinforcement learning (DRL) methods either directly build a separate policy for each type of action or simplify the hybrid action space into a discrete or continuous action control problem. However, these methods neglect the challenge of exploration resulting from the complexity of the hybrid action space. Thus, it is necessary to design more sample efficient algorithms. To this end, we propose a novel Hybrid Control Transformer (Hybrid CtrlFormer), to achieve better exploration and exploitation for the hybrid action control problems. The core idea is: 1) we construct a hybrid action space tree with the discrete actions at the higher level and the continuous parameter space at the lower level. Each parameter space is split into multiple subregions. 2) To simplify the exploration space, a Transformer-based Monte-Carlo tree search method is designed to efficiently evaluate and partition the hybrid action space into good and bad subregions along the tree. Our method achieves state-of-the-art performance and sample efficiency in a variety of environments with discrete-continuous action space.

Individuals have unique facial expression and head pose styles that reflect their personalized speaking styles. Existing one-shot talking head methods cannot capture such personalized characteristics and therefore fail to produce diverse speaking styles in the final videos. To address this challenge, we propose a one-shot style-controllable talking face generation method that can obtain speaking styles from reference speaking videos and drive the one-shot portrait to speak with the reference speaking styles and another piece of audio. Our method aims to synthesize the style-controllable coefficients of a 3D Morphable Model (3DMM), including facial expressions and head movements, in a unified framework. Specifically, the proposed framework first leverages a style encoder to extract the desired speaking styles from the reference videos and transform them into style codes. Then, the framework uses a style-aware decoder to synthesize the coefficients of 3DMM from the audio input and style codes. During decoding, our framework adopts a two-branch architecture, which generates the stylized facial expression coefficients and stylized head movement coefficients, respectively. After obtaining the coefficients of 3DMM, an image renderer renders the expression coefficients into a specific person's talking-head video. Extensive experiments demonstrate that our method generates visually authentic talking head videos with diverse speaking styles from only one portrait image and an audio clip.

As a fine-grained and local expression behavior measurement, facial action unit (FAU) analysis (e.g., detection and intensity estimation) has been documented for its time-consuming, labor-intensive, and error-prone annotation. Thus a long-standing challenge of FAU analysis arises from the data scarcity of manual annotations, limiting the generalization ability of trained models to a large extent. Amounts of previous works have made efforts to alleviate this issue via semi/weakly supervised methods and extra auxiliary information. However, these methods still require domain knowledge and have not yet avoided the high dependency on data annotation. This article introduces a robust facial representation model MAE-Face for AU analysis. Using masked autoencoding as the self-supervised pre-training approach, MAE-Face first learns a high-capacity model from a feasible collection of face images without additional data annotations. Then after being fine-tuned on AU datasets, MAE-Face exhibits convincing performance for both AU detection and AU intensity estimation, achieving a new state-of-the-art on nearly all the evaluation results. Further investigation shows that MAE-Face achieves decent performance even when fine-tuned on only 1% of the AU training set, strongly proving its robustness and generalization performance. The pre-trained model is available at our GitHub repository.

Cheating is widespread in online games, particularly in competitive games like First-Person Shooter (FPS) games. One of the most common types of cheating is Extrasensory Perception (ESP), which involves illicitly obtaining visual information to gain an unfair advantage over normal players. To protect the gaming experience of legitimate players and the interests of game companies, there is an urgent need for anti-cheating applications. In this paper, we propose a general system for ESP anti-cheating in online FPS games, considering the business characteristics and industrial applications. We present a vision-based anti-cheating framework that incorporates both supervised and unsupervised solutions for comprehensive cheating detection. Based on this framework, we design and deploy a dual-audit human-in-the-loop system for industrial gaming anti-cheating applications. We evaluate our proposed framework from multiple online and offline perspectives and demonstrate its practical significance with superior performance.

Massive multiplayer online role-playing games (MMORPG) have been becoming one of the most popular and exciting online games. In recent years, a cheating phenomenon called real money trading (RMT) has arisen and damaged the fantasy world in many ways. RMT is the sale of in-game items, currency, or even characters to earn real money, breaking the balance of the game economy ecosystem and damaging the game experience. Therefore, some studies have emerged to address the problem of RMT detection. However, they cannot well handle the label uncertainty problem in practice, where there are only labeled RMT samples (positive samples) and unlabeled samples, which could either be RMT samples or normal transactions (negative samples). Meanwhile, the trading relationship between RMTers is modeled in a simple way, leading to some normal transactions being falsely classified as RMT. In this article, we propose PU-Detector, a novel framework based on PU learning (learning from positive and unlabeled data) for RMT detection, considering the fact that there are only labeled RMT samples and other unlabeled transactions. We first automatically estimate the likelihood of one transaction being RMT by developing an improved PU learning method and proposing an assessment rule. Sequentially, we use the estimated likelihood as edge weight to construct a trading graph to learn trader representation. Then, with the trader representations and basic trading features, we detect RMT samples by the improved PU learning method. PU-Detector is evaluated on a large-scale real world dataset consisting of 33,809,956 transaction logs generated by 43,217 unique players. Compared with other approaches, it achieves the state-of-the-art performance and demonstrates its advantages in detecting underlying RMT samples.

Different people speak with diverse personalized speaking styles. Although existing one-shot talking head methods have made significant progress in lip sync, natural facial expressions, and stable head motions, they still cannot generate diverse speaking styles in the final talking head videos. To tackle this problem, we propose a one-shot style-controllable talking face generation framework. In a nutshell, we aim to attain a speaking style from an arbitrary reference speaking video and then drive the one-shot portrait to speak with the reference speaking style and another piece of audio. Specifically, we first develop a style encoder to extract dynamic facial motion patterns of a style reference video and then encode them into a style code. Afterward, we introduce a style-controllable decoder to synthesize stylized facial animations from the speech content and style code. In order to integrate the reference speaking style into generated videos, we design a style-aware adaptive transformer, which enables the encoded style code to adjust the weights of the feed-forward layers accordingly. Thanks to the style-aware adaptation mechanism, the reference speaking style can be better embedded into synthesized videos during decoding. Extensive experiments demonstrate that our method is capable of generating talking head videos with diverse speaking styles from only one portrait image and an audio clip while achieving authentic visual effects. Project Page: https://github.com/FuxiVirtualHuman/styletalk.

In this work, we propose a semantic flow-guided two-stage framework for shape-aware face swapping, namely FlowFace. Unlike most previous methods that focus on transferring the source inner facial features but neglect facial contours, our FlowFace can transfer both of them to a target face, thus leading to more realistic face swapping. Concretely, our FlowFace consists of a face reshaping network and a face swapping network. The face reshaping network addresses the shape outline differences between the source and target faces. It first estimates a semantic flow (i.e. face shape differences) between the source and the target face, and then explicitly warps the target face shape with the estimated semantic flow. After reshaping, the face swapping network generates inner facial features that exhibit the identity of the source face. We employ a pre-trained face masked autoencoder (MAE) to extract facial features from both the source face and the target face. In contrast to previous methods that use identity embedding to preserve identity information, the features extracted by our encoder can better capture facial appearances and identity information. Then, we develop a cross-attention fusion module to adaptively fuse inner facial features from the source face with the target facial attributes, thus leading to better identity preservation. Extensive quantitative and qualitative experiments on in-the-wild faces demonstrate that our FlowFace outperforms the state-of-the-art significantly.

For few-shot learning, it is still a critical challenge to realize photo-realistic face visually dubbing on high-resolution videos. Previous works fail to generate high-fidelity dubbing results. To address the above problem, this paper proposes a Deformation Inpainting Network (DINet) for high-resolution face visually dubbing. Different from previous works relying on multiple up-sample layers to directly generate pixels from latent embeddings, DINet performs spatial deformation on feature maps of reference images to better preserve high-frequency textural details. Specifically, DINet consists of one deformation part and one inpainting part. In the first part, five reference facial images adaptively perform spatial deformation to create deformed feature maps encoding mouth shapes at each frame, in order to align with input driving audio and also the head poses of input source images. In the second part, to produce face visually dubbing, a feature decoder is responsible for adaptively incorporating mouth movements from the deformed feature maps and other attributes (i.e., head pose and upper facial expression) from the source feature maps together. Finally, DINet achieves face visually dubbing with rich textural details. We conduct qualitative and quantitative comparisons to validate our DINet on high-resolution videos. The experimental results show that our method outperforms state-of-the-art works.

The use of modern Large Language Models (LLMs) as chatbots still has some problems such as hallucinations and lack of empathy.Identifying these issues can help improve chatbot performance.The community has been continually iterating on reference-free dialogue evaluation methods based on large language models (LLMs) that can be readily applied.However, many of these LLM-based metrics require selecting specific datasets and developing specialized training tasks for different evaluation dimensions (e.g., coherence, informative).The developing step can be time-consuming and may need to be repeated for new evaluation dimensions.To enable efficient and flexible adaptation to diverse needs of dialogue evaluation, we propose a dimension-agnostic scoring method that leverages the in-context learning (ICL) capability of LLMs to learn from human scoring to the fullest extent.Our method has three key features.To begin with, rather than manual prompt crafting, we propose automatically generating prompts, allowing the LLM to observe human labels and summarize the most suitable prompt.Additionally, since the LLM has a token limit and ICL is sensitive to demonstration variations, we train a selector to finely customize demonstrations and prompts for each dialogue input.Finally, during inference, we propose to request the LLM multiple times with a subgraph of demonstrations and prompts that are diverse and suitable to maximize ICL from various human scoring.We validate the efficacy of our method on five datasets, even with a small amount of annotated data, our method outperforms all strong baselines.Code is available at EMNLP2023-ADOROR.

In recent years, 3D representation learning has turned to 2D vision-language pre-trained models to overcome data scarcity challenges. However, existing methods simply transfer 2D alignment strategies, aligning 3D representations with single-view 2D images and coarse-grained parent category text. These approaches introduce information degradation and insufficient synergy issues, leading to performance loss. Information degradation arises from overlooking the fact that a 3D representation should be equivalent to a series of multi-view images and more fine-grained subcategory text. Insufficient synergy neglects the idea that a robust 3D representation should align with the joint vision-language space, rather than independently aligning with each modality. In this paper, we propose a multi-view joint modality modeling approach, termed JM3D, to obtain a unified representation for point cloud, text, and image. Specifically, a novel Structured Multimodal Organizer (SMO) is proposed to address the information degradation issue, which introduces contiguous multi-view images and hierarchical text to enrich the representation of vision and language modalities. A Joint Multi-modal Alignment (JMA) is designed to tackle the insufficient synergy problem, which models the joint modality by incorporating language knowledge into the visual modality. Extensive experiments on ModelNet40 and ScanObjectNN demonstrate the effectiveness of our proposed method, JM3D, which achieves state-of-the-art performance in zero-shot 3D classification. JM3D outperforms ULIP by approximately 4.3% on PointMLP and achieves an improvement of up to 6.5% accuracy on PointNet++ in top-1 accuracy for zero-shot 3D classification on ModelNet40. The source code and trained models for all our experiments are publicly available at https://github.com/Mr-Neko/JM3D.

Prior research on policy adaptation in multi-agent games has often relied on online interaction with the target agent in training, which can be expensive and impractical in real-world scenarios. Inspired by recent progress in offline reinforcement learn- ing, this paper studies offline policy adaptation, which aims to utilize the target agent’s behavior data to exploit its weakness or enable effective cooperation. We investigate its distinct challenges of distributional shift and risk-free deviation, and propose a novel learning objective, conservative offline adaptation, that optimizes the worst-case performance against any dataset consistent proxy models. We pro- pose an efficient algorithm called Constrained Self-Play (CSP) that incorporates dataset information into regularized policy learning. We prove that CSP learns a near-optimal risk-free offline adaptation policy upon convergence. Empirical results demonstrate that CSP outperforms non-conservative baselines in various environments, including Maze, predator-prey, MuJoCo, and Google Football.

As an independent social and economic entity, game servers plays a dominant role in building a stable, living, and attractive virtual world in massive multi-player online role-playing games (MMORPGs). We propose and implement a novel intelligent decision support system for server merge (SM) for maintaining the game ecology at the macro level. The services provided by this system include server health diagnosis, server merge assessment, and combination strategy recommendation. Specifically, we design an effective time series prediction algorithm to diagnose the health status of one server (e.g., user activity, online time, daily revenue) based on real game scenarios, and then select the servers with poor status from all servers. Moreover, to dig out the inherent development laws of servers from the historical merge records, we leverage a correlation measurement algorithm to find the historical merged servers that are similar to the servers to be merged and then evaluate the potential trend after merging, which can assist experts to make reasonable decisions. We deploy our system into practice for multiple MMORPGs and achieve sound online performance endorsed by the game operation team.

In many real-world multi-agent systems, the sparsity of team rewards often makes it difficult for an algorithm to successfully learn a cooperative team policy. At present, the common way for solving this problem is to design some dense individual rewards for the agents to guide the cooperation. However, most existing works utilize individual rewards in ways that do not always promote teamwork and sometimes are even counterproductive. In this paper, we propose Individual Reward Assisted Team Policy Learning (IRAT), which learns two policies for each agent from the dense individual reward and the sparse team reward with discrepancy constraints for updating the two policies mutually. Experimental results in different scenarios, such as the Multi-Agent Particle Environment and the Google Research Football Environment, show that IRAT significantly outperforms the baseline methods and can greatly promote team policy learning without deviating from the original team objective, even when the individual rewards are misleading or conflict with the team rewards.

Recent years have witnessed the great accuracy performance of graph-based Collaborative Filtering (CF) models for recommender systems. By taking the user-item interaction behavior as a graph, these graph-based CF models borrow the success of Graph Neural Networks (GNN), and iteratively perform neighborhood aggregation to propagate the collaborative signals. While conventional CF models are known for facing the challenges of the popularity bias that favors popular items, one may wonder "Whether the existing graph-based CF models alleviate or exacerbate the popularity bias of recommender systems?" To answer this question, we first investigate the two-fold performances w.r.t. accuracy and novelty for existing graph-based CF methods. The empirical results show that symmetric neighborhood aggregation adopted by most existing graph-based CF models exacerbates the popularity bias and this phenomenon becomes more serious as the depth of graph propagation increases. Further, we theoretically analyze the cause of popularity bias for graph-based CF. Then, we propose a simple yet effective plugin, namely r-AdjNorm, to achieve an accuracy-novelty trade-off by controlling the normalization strength in the neighborhood aggregation process. Meanwhile, r-AdjNorm can be smoothly applied to the existing graph-based CF backbones without additional computation. Finally, experimental results on three benchmark datasets show that our proposed method can improve novelty without sacrificing accuracy under various graph-based CF backbones.

It is intuitive that different actions prefer different information in human decisions. However, classical reinforcement learning models use the same information process procedure for all actions. In order to imitate human decision-making process closer, in this paper we investigate a new policy model, i.e., Action-Specific Focuses (ASF) framework, which enables different focuses when learning different actions. In the ASF framework, the whole action set is taken as part of the queries for the attention module, in which state-dependent action-specific features can be generated. Through extracting different action-specific features, our approach enables the agent to learn the action-focus map for each action separately. The ASF framework is also different from the previous usages of attention mechanisms in reinforcement learning that are mostly based on the state. Experiments on the Atari benchmark show that ASF is able to improve the performance in various types of games. Moreover, the visualizations of the attention weights suggest that ASF can learn meaningful focuses when taking different actions.

Experience reuse is key to sample-efficient reinforcement learning. One of the critical issues is how the experience is represented and stored. Previously, the experience can be stored in the forms of features, individual models, and the average model, each lying at a different granularity. However, new tasks may require experience across multiple granularities. In this paper, we propose the policy residual representation (PRR) network, which can extract and store multiple levels of experience. PRR network is trained on a set of tasks with a multi-level architecture, where a module in each level corresponds to a subset of the tasks. Therefore, the PRR network represents the experience in a spectrum-like way. When training on a new task, PRR can provide different levels of experience for accelerating the learning. We experiment with the PRR network on a set of grid world navigation tasks, locomotion tasks, and fighting tasks in a video game. The results show that the PRR network leads to better reuse of experience and thus outperforms some state-of-the-art approaches.