Search

The advancement of Offline Reinforcement Learning (RL) and Offline Multi-Agent Reinforcement Learning (MARL) critically depends on the availability of high-quality, pre-collected offline datasets that represent real-world complexities and practical applications. However, existing datasets often fall short in their simplicity and lack of realism. To address this gap, we propose Hokoff, a comprehensive set of pre-collected datasets that covers both offline RL and offline MARL, accompanied by a robust framework, to facilitate further research. This data is derived from Honor of Kings, a recognized Multiplayer Online Battle Arena (MOBA) game known for its intricate nature, closely resembling real-life situations. Utilizing this framework, we benchmark a variety of offline RL and offline MARL algorithms. We also introduce a novel baseline algorithm tailored for the inherent hierarchical action space of the game. We reveal the incompetency of current offline RL approaches in handling task complexity, generalization and multi-task learning.

Digital imaging systems have classically been based on brute-force measuring and processing of pixels organized on regular grids. The human visual system, on the other hand, performs a massive data reduction from the number of photo-receptors to the optic nerve, essentially encoding the image information into a low bandwidth latent space representation suitable for processing by the human brain. In this work, we propose to follow a similar approach for the development of artificial vision systems. Latent Space Imaging is a new paradigm that, through a combination of optics and software, directly encodes the image information into the semantically rich latent space of a generative model, thus substantially reducing bandwidth and memory requirements during the capture process. We demonstrate this new principle through an initial hardware prototype based on the single pixel camera. By designing an amplitude modulation scheme that encodes into the latent space of a generative model, we achieve compression ratios from 1:100 to 1:1,000 during the imaging process, illustrating the potential of latent space imaging for highly efficient imaging hardware, to enable future applications in high speed imaging, or task-specific cameras with substantially reduced hardware complexity.

Recent years, Pre-trained Language models (PLMs) have swept into various fields of artificial intelligence and achieved great success. However, most PLMs, such as T5 and GPT3, have a huge amount of parameters, fine-tuning them is often expensive and time consuming, and storing them takes up a lot of space. Therefore, it is necessary to adopt a parameter-efficient approach to reduce parameters of PLMs in fine-tuning without compromising their performance in downstream tasks. In this paper, we design a novel adapter which only acts on self-attention outputs in PLMs. This adapter adopts element-wise linear transformation using Hadamard product, hence named as Hadamard adapter, requires the fewest parameters compared to previous parameter-efficient adapters. In addition, we also summarize some tuning patterns for Hadamard adapter shared by various downstream tasks, expecting to provide some guidance for further parameter reduction with shared adapters in future studies. The experiments conducted on the widely-used GLUE benchmark with several SOTA PLMs prove that the Hadamard adapter achieves competitive performance with only 0.033\% parameters compared with full fine-tuning, and it has the fewest parameters compared with other adapters. Moreover, we further find that there is also some redundant layers in the Hadamard adapter which can be removed to achieve more parameter efficiency with only 0.022\% parameters., Comment: Accepted to CIKM 2023 (Long Paper)

Large Language Models (LLMs) have gained widespread adoption in various natural language processing tasks, including question answering and dialogue systems. However, a major drawback of LLMs is the issue of hallucination, where they generate unfaithful or inconsistent content that deviates from the input source, leading to severe consequences. In this paper, we propose a robust discriminator named RelD to effectively detect hallucination in LLMs' generated answers. RelD is trained on the constructed RelQA, a bilingual question-answering dialogue dataset along with answers generated by LLMs and a comprehensive set of metrics. Our experimental results demonstrate that the proposed RelD successfully detects hallucination in the answers generated by diverse LLMs. Moreover, it performs well in distinguishing hallucination in LLMs' generated answers from both in-distribution and out-of-distribution datasets. Additionally, we also conduct a thorough analysis of the types of hallucinations that occur and present valuable insights. This research significantly contributes to the detection of reliable answers generated by LLMs and holds noteworthy implications for mitigating hallucination in the future work., Comment: Accepted to CIKM 2023 (Long Paper)

Differentiable causal discovery has made significant advancements in the learning of directed acyclic graphs. However, its application to real-world datasets remains restricted due to the ubiquity of latent confounders and the requirement to learn maximal ancestral graphs (MAGs). To date, existing differentiable MAG learning algorithms have been limited to small datasets and failed to scale to larger ones (e.g., with more than 50 variables). The key insight in this paper is that the causal skeleton, which is the undirected version of the causal graph, has potential for improving accuracy and reducing the search space of the optimization procedure, thereby enhancing the performance of differentiable causal discovery. Therefore, we seek to address a two-fold challenge to harness the potential of the causal skeleton for differentiable causal discovery in the presence of latent confounders: (1) scalable and accurate estimation of skeleton and (2) universal integration of skeleton estimation with differentiable causal discovery. To this end, we propose SPOT (Skeleton Posterior-guided OpTimization), a two-phase framework that harnesses skeleton posterior for differentiable causal discovery in the presence of latent confounders. On the contrary to a ``point-estimation'', SPOT seeks to estimate the posterior distribution of skeletons given the dataset. It first formulates the posterior inference as an instance of amortized inference problem and concretizes it with a supervised causal learning (SCL)-enabled solution to estimate the skeleton posterior. To incorporate the skeleton posterior with differentiable causal discovery, SPOT then features a skeleton posterior-guided stochastic optimization procedure to guide the optimization of MAGs. [abridged due to length limit]

Hybrid refractive-diffractive lenses combine the light efficiency of refractive lenses with the information encoding power of diffractive optical elements (DOE), showing great potential as the next generation of imaging systems. However, accurately simulating such hybrid designs is generally difficult, and in particular, there are no existing differentiable image formation models for hybrid lenses with sufficient accuracy. In this work, we propose a new hybrid ray-tracing and wave-propagation (ray-wave) model for accurate simulation of both optical aberrations and diffractive phase modulation, where the DOE is placed between the last refractive surface and the image sensor, i.e. away from the Fourier plane that is often used as a DOE position. The proposed ray-wave model is fully differentiable, enabling gradient back-propagation for end-to-end co-design of refractive-diffractive lens optimization and the image reconstruction network. We validate the accuracy of the proposed model by comparing the simulated point spread functions (PSFs) with theoretical results, as well as simulation experiments that show our model to be more accurate than solutions implemented in commercial software packages like Zemax. We demonstrate the effectiveness of the proposed model through real-world experiments and show significant improvements in both aberration correction and extended depth-of-field (EDoF) imaging. We believe the proposed model will motivate further investigation into a wide range of applications in computational imaging, computational photography, and advanced optical design. Code will be released upon publication.

Counterfactual regret minimization (CFR) is a family of algorithms for effectively solving imperfect-information games. It decomposes the total regret into counterfactual regrets, utilizing local regret minimization algorithms, such as Regret Matching (RM) or RM+, to minimize them. Recent research establishes a connection between Online Mirror Descent (OMD) and RM+, paving the way for an optimistic variant PRM+ and its extension PCFR+. However, PCFR+ assigns uniform weights for each iteration when determining regrets, leading to substantial regrets when facing dominated actions. This work explores minimizing weighted counterfactual regret with optimistic OMD, resulting in a novel CFR variant PDCFR+. It integrates PCFR+ and Discounted CFR (DCFR) in a principled manner, swiftly mitigating negative effects of dominated actions and consistently leveraging predictions to accelerate convergence. Theoretical analyses prove that PDCFR+ converges to a Nash equilibrium, particularly under distinct weighting schemes for regrets and average strategies. Experimental results demonstrate PDCFR+'s fast convergence in common imperfect-information games. The code is available at https://github.com/rpSebastian/PDCFRPlus., Comment: Accepted to 33rd International Joint Conference on Artificial Intelligence (IJCAI 2024)

X-ray absorption spectroscopy (XAS) is widely employed for structure characterization of graphitic carbon nitride (g-C$_3$N$_4$) and its composites. Nevertheless, even for pure g-C$_3$N$_4$, discrepancies in energy and profile exist across different experiments, which can be attributed to variations in structures arising from diverse synthesis conditions and calibration procedures. Here, we conducted a theoretical investigation on XAS of three representative g-C$_3$N$_4$ structures (planar, corrugated, and micro-corrugated) optimized with different strategies, to understand the structure-spectroscopy relation. Different methods were compared, including density functional theory (DFT) with the full (FCH) or equivalent (ECH) core-hole approximation, as well as the time-dependent DFT (TDDFT). FCH was responsible for getting accurate absolute absorption energy; while ECH and TDDFT aided in interpreting the spectra, through ECH-state canonical molecular orbitals (ECH-CMOs) and natural transition orbitals (NTOs), respectively. With each method, the spectra at the three structures show evident differences, which can be correlated to different individual experiments or in between. Our calculations explained the structural reason behind the spectral discrepancies among different experiments. Moreover, profiles predicted by these methods also displayed consistency, so their differences can be used as a reliable indicator of their accuracy. Both ECH-CMOs and NTO particle orbitals led to similar graphics, validating their applicability in interpreting the transitions. This work provides a comprehensive analysis of the structure-XAS relation for g-C$_3$N$_4$, provides concrete explanations for the spectral differences reported in various experiments, and offers insight for future structure dynamical and transient X-ray spectral analyses., Comment: 6 pages, 4 figures

Reaching consensus is key to multi-agent coordination. To accomplish a cooperative task, agents need to coherently select optimal joint actions to maximize the team reward. However, current cooperative multi-agent reinforcement learning (MARL) methods usually do not explicitly take consensus into consideration, which may cause miscoordination problem. In this paper, we propose a model-based consensus mechanism to explicitly coordinate multiple agents. The proposed Multi-agent Goal Imagination (MAGI) framework guides agents to reach consensus with an Imagined common goal. The common goal is an achievable state with high value, which is obtained by sampling from the distribution of future states. We directly model this distribution with a self-supervised generative model, thus alleviating the "curse of dimensinality" problem induced by multi-agent multi-step policy rollout commonly used in model-based methods. We show that such efficient consensus mechanism can guide all agents cooperatively reaching valuable future states. Results on Multi-agent Particle-Environments and Google Research Football environment demonstrate the superiority of MAGI in both sample efficiency and performance.

In recent years, neural network-based Wake Word Spotting achieves good performance on clean audio samples but struggles in noisy environments. Audio-Visual Wake Word Spotting (AVWWS) receives lots of attention because visual lip movement information is not affected by complex acoustic scenes. Previous works usually use simple addition or concatenation for multi-modal fusion. The inter-modal correlation remains relatively under-explored. In this paper, we propose a novel module called Frame-Level Cross-Modal Attention (FLCMA) to improve the performance of AVWWS systems. This module can help model multi-modal information at the frame-level through synchronous lip movements and speech signals. We train the end-to-end FLCMA based Audio-Visual Conformer and further improve the performance by fine-tuning pre-trained uni-modal models for the AVWWS task. The proposed system achieves a new state-of-the-art result (4.57% WWS score) on the far-field MISP dataset., Comment: Accepted by ICASSP 2024

Foundation models, such as Large language Models (LLMs), have attracted significant amount of interest due to their large number of applications. However, when handling tasks involving repetitive sub-tasks and/or deceptive contents, such as arithmetic calculation and article-level fake news detection, simple instructional prompts suffer from inaccurate responses. Existing works show that more complicated prompting strategies, such as Chain-of-Thoughts and Least-to-Most, can unlock LLM's powerful capacity in diverse areas. Recent researches reveal that simple divide-and-conquer prompting strategy, i.e. simply dividing the input sequence to multiple sub-inputs, can also substantially improve LLM's performance in some specific tasks such as misinformation detection. In this paper, we aim at examining the utility of divide-and-conquer prompting strategy and answer on which kind of tasks this strategy gets advantages. Specifically, we provide a theoretic analysis to divide-and-conquer prompting strategy and help us identify the specific tasks where DaC prompting can bring performance boost with theoretic guarantee. We then present two cases (large integer arithmetic and fact verification) where experimental results aligns with our theoretic analysis., Comment: Preprint

This paper presents an innovative framework that integrates Large Language Models (LLMs) with an external Thinker module to enhance the reasoning capabilities of LLM-based agents. Unlike augmenting LLMs with prompt engineering, Thinker directly harnesses knowledge from databases and employs various optimization techniques. The framework forms a reasoning hierarchy where LLMs handle intuitive System-1 tasks such as natural language processing, while the Thinker focuses on cognitive System-2 tasks that require complex logical analysis and domain-specific knowledge. Our framework is presented using a 9-player Werewolf game that demands dual-system reasoning. We introduce a communication protocol between LLMs and the Thinker, and train the Thinker using data from 18800 human sessions and reinforcement learning. Experiments demonstrate the framework's effectiveness in deductive reasoning, speech generation, and online game evaluation. Additionally, we fine-tune a 6B LLM to surpass GPT4 when integrated with the Thinker. This paper also contributes the largest dataset for social deduction games to date.

We find that, simply via a sampling-and-voting method, the performance of large language models (LLMs) scales with the number of agents instantiated. Also, this method is orthogonal to existing complicated methods to further enhance LLMs, while the degree of enhancement is correlated to the task difficulty. We conduct comprehensive experiments on a wide range of LLM benchmarks to verify the presence of our finding, and to study the properties that can facilitate its occurrence. Our code is publicly available at: \url{https://anonymous.4open.science/r/more_agent_is_all_you_need}.

The emergence of large language models (LLMs) has significantly advanced the simulation of believable interactive agents. However, the substantial cost on maintaining the prolonged agent interactions poses challenge over the deployment of believable LLM-based agents. Therefore, in this paper, we develop Affordable Generative Agents (AGA), a framework for enabling the generation of believable and low-cost interactions on both agent-environment and inter-agents levels. Specifically, for agent-environment interactions, we substitute repetitive LLM inferences with learned policies; while for inter-agent interactions, we model the social relationships between agents and compress auxiliary dialogue information. Extensive experiments on multiple environments show the effectiveness and efficiency of our proposed framework. Also, we delve into the mechanisms of emergent believable behaviors lying in LLM agents, demonstrating that agents can only generate finite behaviors in fixed environments, based upon which, we understand ways to facilitate emergent interaction behaviors. Our code is publicly available at: https://github.com/AffordableGenerativeAgents/Affordable-Generative-Agents.

Existing game AI research mainly focuses on enhancing agents' abilities to win games, but this does not inherently make humans have a better experience when collaborating with these agents. For example, agents may dominate the collaboration and exhibit unintended or detrimental behaviors, leading to poor experiences for their human partners. In other words, most game AI agents are modeled in a "self-centered" manner. In this paper, we propose a "human-centered" modeling scheme for collaborative agents that aims to enhance the experience of humans. Specifically, we model the experience of humans as the goals they expect to achieve during the task. We expect that agents should learn to enhance the extent to which humans achieve these goals while maintaining agents' original abilities (e.g., winning games). To achieve this, we propose the Reinforcement Learning from Human Gain (RLHG) approach. The RLHG approach introduces a "baseline", which corresponds to the extent to which humans primitively achieve their goals, and encourages agents to learn behaviors that can effectively enhance humans in achieving their goals better. We evaluate the RLHG agent in the popular Multi-player Online Battle Arena (MOBA) game, Honor of Kings, by conducting real-world human-agent tests. Both objective performance and subjective preference results show that the RLHG agent provides participants better gaming experience., Comment: Accepted at ICLR 2024. arXiv admin note: text overlap with arXiv:2304.11632

Hyperspectral imaging empowers machine vision systems with the distinct capability of identifying materials through recording their spectral signatures. Recent efforts in data-driven spectral reconstruction aim at extracting spectral information from RGB images captured by cost-effective RGB cameras, instead of dedicated hardware. In this paper we systematically analyze the performance of such methods, evaluating both the practical limitations with respect to current datasets and overfitting, as well as fundamental limitations with respect to the nature of the information encoded in the RGB images, and the dependency of this information on the optical system of the camera. We find that, the current models are not robust under slight variations, e.g., in noise level or compression of the RGB file. Without modeling underrepresented spectral content, existing datasets and the models trained on them are limited in their ability to cope with challenging metameric colors. To mitigate this issue, we propose to exploit the combination of metameric data augmentation and optical lens aberrations to improve the encoding of the metameric information into the RGB image, which paves the road towards higher performing spectral imaging and reconstruction approaches., Comment: 12 pages, 7 figures, 8 tables

We propose a new method called the N-particle underdamped Langevin algorithm for optimizing a special class of non-linear functionals defined over the space of probability measures. Examples of problems with this formulation include training mean-field neural networks, maximum mean discrepancy minimization and kernel Stein discrepancy minimization. Our algorithm is based on a novel spacetime discretization of the mean-field underdamped Langevin dynamics, for which we provide a new, fast mixing guarantee. In addition, we demonstrate that our algorithm converges globally in total variation distance, bridging the theoretical gap between the dynamics and its practical implementation., Comment: 40 pages, 5 figures, 2 tables

The rapid advancements in artificial intelligence (AI) are catalyzing transformative changes in atomic modeling, simulation, and design. AI-driven potential energy models have demonstrated the capability to conduct large-scale, long-duration simulations with the accuracy of ab initio electronic structure methods. However, the model generation process remains a bottleneck for large-scale applications. We propose a shift towards a model-centric ecosystem, wherein a large atomic model (LAM), pre-trained across multiple disciplines, can be efficiently fine-tuned and distilled for various downstream tasks, thereby establishing a new framework for molecular modeling. In this study, we introduce the DPA-2 architecture as a prototype for LAMs. Pre-trained on a diverse array of chemical and materials systems using a multi-task approach, DPA-2 demonstrates superior generalization capabilities across multiple downstream tasks compared to the traditional single-task pre-training and fine-tuning methodologies. Our approach sets the stage for the development and broad application of LAMs in molecular and materials simulation research.

Multi-task reinforcement learning endeavors to accomplish a set of different tasks with a single policy. To enhance data efficiency by sharing parameters across multiple tasks, a common practice segments the network into distinct modules and trains a routing network to recombine these modules into task-specific policies. However, existing routing approaches employ a fixed number of modules for all tasks, neglecting that tasks with varying difficulties commonly require varying amounts of knowledge. This work presents a Dynamic Depth Routing (D2R) framework, which learns strategic skipping of certain intermediate modules, thereby flexibly choosing different numbers of modules for each task. Under this framework, we further introduce a ResRouting method to address the issue of disparate routing paths between behavior and target policies during off-policy training. In addition, we design an automatic route-balancing mechanism to encourage continued routing exploration for unmastered tasks without disturbing the routing of mastered ones. We conduct extensive experiments on various robotics manipulation tasks in the Meta-World benchmark, where D2R achieves state-of-the-art performance with significantly improved learning efficiency., Comment: AAAI2024, with supplementary material

Online recruitment platforms typically employ Person-Job Fit models in the core service that automatically match suitable job seekers with appropriate job positions. While existing works leverage historical or contextual information, they often disregard a crucial aspect: job seekers' social relationships in professional networks. This paper emphasizes the importance of incorporating professional networks into the Person-Job Fit model. Our innovative approach consists of two stages: (1) defining a Workplace Heterogeneous Information Network (WHIN) to capture heterogeneous knowledge, including professional connections and pre-training representations of various entities using a heterogeneous graph neural network; (2) designing a Contextual Social Attention Graph Neural Network (CSAGNN) that supplements users' missing information with professional connections' contextual information. We introduce a job-specific attention mechanism in CSAGNN to handle noisy professional networks, leveraging pre-trained entity representations from WHIN. We demonstrate the effectiveness of our approach through experimental evaluations conducted across three real-world recruitment datasets from LinkedIn, showing superior performance compared to baseline models., Comment: Accepted at WSDM 2024

Achieving high performance for Sparse MatrixMatrix Multiplication (SpMM) has received increasing research attention, especially on multi-core CPUs, due to the large input data size in applications such as graph neural networks (GNNs). Most existing solutions for SpMM computation follow the aheadof-time (AOT) compilation approach, which compiles a program entirely before it is executed. AOT compilation for SpMM faces three key limitations: unnecessary memory access, additional branch overhead, and redundant instructions. These limitations stem from the fact that crucial information pertaining to SpMM is not known until runtime. In this paper, we propose JITSPMM, a just-in-time (JIT) assembly code generation framework to accelerated SpMM computation on multi-core CPUs with SIMD extensions. First, JITSPMM integrates the JIT assembly code generation technique into three widely-used workload division methods for SpMM to achieve balanced workload distribution among CPU threads. Next, with the availability of runtime information, JITSPMM employs a novel technique, coarse-grain column merging, to maximize instruction-level parallelism by unrolling the performance-critical loop. Furthermore, JITSPMM intelligently allocates registers to cache frequently accessed data to minimizing memory accesses, and employs selected SIMD instructions to enhance arithmetic throughput. We conduct a performance evaluation of JITSPMM and compare it two AOT baselines. The first involves existing SpMM implementations compiled using the Intel icc compiler with auto-vectorization. The second utilizes the highly-optimized SpMM routine provided by Intel MKL. Our results show that JITSPMM provides an average improvement of 3.8x and 1.4x, respectively.

Diversity plays a significant role in many problems, such as ensemble learning, reinforcement learning, and combinatorial optimization. How to define the diversity measure is a longstanding problem. Many methods rely on expert experience to define a proper behavior space and then obtain the diversity measure, which is, however, challenging in many scenarios. In this paper, we propose the problem of learning a behavior space from human feedback and present a general method called Diversity from Human Feedback (DivHF) to solve it. DivHF learns a behavior descriptor consistent with human preference by querying human feedback. The learned behavior descriptor can be combined with any distance measure to define a diversity measure. We demonstrate the effectiveness of DivHF by integrating it with the Quality-Diversity optimization algorithm MAP-Elites and conducting experiments on the QDax suite. The results show that DivHF learns a behavior space that aligns better with human requirements compared to direct data-driven approaches and leads to more diverse solutions under human preference. Our contributions include formulating the problem, proposing the DivHF method, and demonstrating its effectiveness through experiments.

Recent years have witnessed the substantial progress of large-scale models across various domains, such as natural language processing and computer vision, facilitating the expression of concrete concepts. Unlike concrete concepts that are usually directly associated with physical objects, expressing abstract concepts through natural language requires considerable effort, which results from their intricate semantics and connotations. An alternative approach is to leverage images to convey rich visual information as a supplement. Nevertheless, existing Text-to-Image (T2I) models are primarily trained on concrete physical objects and tend to fail to visualize abstract concepts. Inspired by the three-layer artwork theory that identifies critical factors, intent, object and form during artistic creation, we propose a framework of Text-to-Image generation for Abstract Concepts (TIAC). The abstract concept is clarified into a clear intent with a detailed definition to avoid ambiguity. LLMs then transform it into semantic-related physical objects, and the concept-dependent form is retrieved from an LLM-extracted form pattern set. Information from these three aspects will be integrated to generate prompts for T2I models via LLM. Evaluation results from human assessments and our newly designed metric concept score demonstrate the effectiveness of our framework in creating images that can sufficiently express abstract concepts.

ERA5 data for the period of 1979-2019 are used as a basis for investigating the properties of gravity waves as they disperse and propagate upward through the stratosphere during opposing phases of the QBO. Two-sided zonal wavenumber-frequency spectra of vertical velocity in the stratosphere exhibit distinctive gravity wave signatures. Consistent with theory, westward propagating waves tend to be suppressed during the easterly QBO phase and eastward propagating waves tend to be suppressed during the westerly phase. Cospectra of the vertical flux of zonal momentum also show significant asymmetries between eastward and westward propagating waves during opposing QBO phases. Phase speed spectra of the vertical flux of momentum are indicative of a strong dissipation of westward propagating gravity waves during the easterly phase and vice versa; i.e., a selective "wind filtering" of the waves as they propagate upward. The three-dimensional structure of the gravity waves is revealed by compositing. In the absence of a background zonal flow, the waves radiate outward and upward from their respective reference grid points in concentric rings. When a zonal flow is present, the rings are amplified and compressed upstream of the source and attenuated and stretched downstream of it, such that they assume the form of arcs. These results serve to confirm the applicability of the mechanism proposed by Lindzen and Holton (1968) to explain the downward propagation of the QBO. The QBO also influences the spectrum of gravity waves in ERA5 at 100 hPa, below the layer in which wind-filtering occurs.

Most existing 3D instance segmentation methods are derived from 3D semantic segmentation models. However, these indirect approaches suffer from certain limitations. They fail to fully leverage global and local semantic information for accurate prediction, which hampers the overall performance of the 3D instance segmentation framework. To address these issues, this paper presents PSGformer, a novel 3D instance segmentation network. PSGformer incorporates two key advancements to enhance the performance of 3D instance segmentation. Firstly, we propose a Multi-Level Semantic Aggregation Module, which effectively captures scene features by employing foreground point filtering and multi-radius aggregation. This module enables the acquisition of more detailed semantic information from global and local perspectives. Secondly, PSGformer introduces a Parallel Feature Fusion Transformer Module that independently processes super-point features and aggregated features using transformers. The model achieves a more comprehensive feature representation by the features which connect global and local features. We conducted extensive experiments on the ScanNetv2 dataset. Notably, PSGformer exceeds compared state-of-the-art methods by 2.2% on ScanNetv2 hidden test set in terms of mAP. Our code and models will be publicly released.

The goal of program synthesis, or code generation, is to generate executable code based on given descriptions. Recently, there has been an increasing number of studies employing reinforcement learning (RL) to improve the performance of large language models (LLMs) for code. However, current representative works either rely solely on offline frameworks, limiting the exploration of new sample spaces, or fall short in the utilization of unit test signals, not accounting for specific error locations within the code. To address these issues, we propose RLTF, i.e., Reinforcement Learning from Unit Test Feedback, a novel online RL framework with unit test feedback of multi-granularity for refining code LLMs. Our approach generates data in real-time during training and simultaneously utilizes fine-grained feedback signals to guide the model towards producing higher-quality code. Extensive experiments show that RLTF achieves state-of-the-art performance on the APPS and the MBPP benchmarks. Our code is available at: https://github.com/Zyq-scut/RLTF., Comment: Accepted by TMLR

Policy-Space Response Oracles (PSRO) is an influential algorithm framework for approximating a Nash Equilibrium (NE) in multi-agent non-transitive games. Many previous studies have been trying to promote policy diversity in PSRO. A major weakness in existing diversity metrics is that a more diverse (according to their diversity metrics) population does not necessarily mean (as we proved in the paper) a better approximation to a NE. To alleviate this problem, we propose a new diversity metric, the improvement of which guarantees a better approximation to a NE. Meanwhile, we develop a practical and well-justified method to optimize our diversity metric using only state-action samples. By incorporating our diversity regularization into the best response solving in PSRO, we obtain a new PSRO variant, Policy Space Diversity PSRO (PSD-PSRO). We present the convergence property of PSD-PSRO. Empirically, extensive experiments on various games demonstrate that PSD-PSRO is more effective in producing significantly less exploitable policies than state-of-the-art PSRO variants.

Multi-agent reinforcement learning (MARL) has been shown effective for cooperative games in recent years. However, existing state-of-the-art methods face challenges related to sample complexity, training instability, and the risk of converging to a suboptimal Nash Equilibrium. In this paper, we propose a unified framework for learning \emph{stochastic} policies to resolve these issues. We embed cooperative MARL problems into probabilistic graphical models, from which we derive the maximum entropy (MaxEnt) objective for MARL. Based on the MaxEnt framework, we propose Heterogeneous-Agent Soft Actor-Critic (HASAC) algorithm. Theoretically, we prove the monotonic improvement and convergence to quantal response equilibrium (QRE) properties of HASAC. Furthermore, we generalize a unified template for MaxEnt algorithmic design named Maximum Entropy Heterogeneous-Agent Mirror Learning (MEHAML), which provides any induced method with the same guarantees as HASAC. We evaluate HASAC on six benchmarks: Bi-DexHands, Multi-Agent MuJoCo, StarCraft Multi-Agent Challenge, Google Research Football, Multi-Agent Particle Environment, and Light Aircraft Game. Results show that HASAC consistently outperforms strong baselines, exhibiting better sample efficiency, robustness, and sufficient exploration., Comment: ICLR 2024 spotlight

Collaborative filtering (CF) is an important research direction in recommender systems that aims to make recommendations given the information on user-item interactions. Graph CF has attracted more and more attention in recent years due to its effectiveness in leveraging high-order information in the user-item bipartite graph for better recommendations. Specifically, recent studies show the success of graph neural networks (GNN) for CF is attributed to its low-pass filtering effects. However, current researches lack a study of how different signal components contributes to recommendations, and how to design strategies to properly use them well. To this end, from the view of spectral transformation, we analyze the important factors that a graph filter should consider to achieve better performance. Based on the discoveries, we design JGCF, an efficient and effective method for CF based on Jacobi polynomial bases and frequency decomposition strategies. Extensive experiments on four widely used public datasets show the effectiveness and efficiency of the proposed methods, which brings at most 27.06% performance gain on Alibaba-iFashion. Besides, the experimental results also show that JGCF is better at handling sparse datasets, which shows potential in making recommendations for cold-start users.

The study of hadronic structure has been carried out for many years. Generalized parton distribution functions (GPDs) give broad information on the internal structure of hadrons. Combining GPDs and high-energy scattering experiments, we expect yielding three-dimensional physical quantities from experiments. Deeply Virtual Compton Scattering (DVCS) process is a powerful tool to study GPDs. It is one of the important experiments of Electron Ion Collider (EIC) and Electron ion collider at China (EicC) in the future. In the initial stage, the proposed EicC will have $3 \sim 5$ GeV polarized electrons on $12 \sim 25$ GeV polarized protons, with luminosity up to $1 \sim 2 \times 10^{33}$cm$^{-2}$s$^{-1}$. EIC will be constructed in coming years, which will cover the variable c.m. energies from 30 to 50 GeV, with the luminosity about $10^{33} \sim 10^{34}$cm$^{-2}$s$^{-1}$. In this work we present a detailed simulation of DVCS to study the feasibility of experiments at EicC and EIC. Referring the method used by HERMES Collaboration, and comparing the model calculations with pseudo data of asymmetries attributed to the DVCS, we obtained a model-dependent constraint on the total angular momentum of up and down quarks in the proton., Comment: 12 pages, 18 figures, 3 Tables

In computer vision, it has long been taken for granted that high-quality images obtained through well-designed camera lenses would lead to superior results. However, we find that this common perception is not a "one-size-fits-all" solution for diverse computer vision tasks. We demonstrate that task-driven and deep-learned simple optics can actually deliver better visual task performance. The Task-Driven lens design approach, which relies solely on a well-trained network model for supervision, is proven to be capable of designing lenses from scratch. Experimental results demonstrate the designed image classification lens (``TaskLens'') exhibits higher accuracy compared to conventional imaging-driven lenses, even with fewer lens elements. Furthermore, we show that our TaskLens is compatible with various network models while maintaining enhanced classification accuracy. We propose that TaskLens holds significant potential, particularly when physical dimensions and cost are severely constrained., Comment: Use an image classification network to supervise the lens design from scratch. The final designs can achieve higher accuracy with fewer optical elements

Recent research in offline reinforcement learning (RL) has demonstrated that return-conditioned supervised learning is a powerful paradigm for decision-making problems. While promising, return conditioning is limited to training data labeled with rewards and therefore faces challenges in learning from unsupervised data. In this work, we aim to utilize generalized future conditioning to enable efficient unsupervised pretraining from reward-free and sub-optimal offline data. We propose Pretrained Decision Transformer (PDT), a conceptually simple approach for unsupervised RL pretraining. PDT leverages future trajectory information as a privileged context to predict actions during training. The ability to make decisions based on both present and future factors enhances PDT's capability for generalization. Besides, this feature can be easily incorporated into a return-conditioned framework for online finetuning, by assigning return values to possible futures and sampling future embeddings based on their respective values. Empirically, PDT outperforms or performs on par with its supervised pretraining counterpart, especially when dealing with sub-optimal data. Further analysis reveals that PDT can extract diverse behaviors from offline data and controllably sample high-return behaviors by online finetuning. Code is available at here., Comment: 17 pages, 9 figures, ICML 2023

In-context learning is the paradigm that adapts large language models to downstream tasks by providing a few examples. Few-shot selection -- selecting appropriate examples for each test instance separately -- is important for in-context learning. In this paper, we propose Skill-KNN, a skill-based few-shot selection method for in-context learning. The key advantages of Skill-KNN include: (1) it addresses the problem that existing methods based on pre-trained embeddings can be easily biased by surface natural language features that are not important for the target task; (2) it does not require training or fine-tuning of any models, making it suitable for frequently expanding or changing example banks. The key insight is to optimize the inputs fed into the embedding model, rather than tuning the model itself. Technically, Skill-KNN generates the skill-based descriptions for each test case and candidate example by utilizing a pre-processing few-shot prompting, thus eliminating unimportant surface features. Experimental results across five cross-domain semantic parsing datasets and six backbone models show that Skill-KNN significantly outperforms existing methods., Comment: Accepted by EMNLP 2023 main conference

Recent years have witnessed the great potential of attention mechanism in graph representation learning. However, while variants of attention-based GNNs are setting new benchmarks for numerous real-world datasets, recent works have pointed out that their induced attentions are less robust and generalizable against noisy graphs due to lack of direct supervision. In this paper, we present a new framework which utilizes the tool of causality to provide a powerful supervision signal for the learning process of attention functions. Specifically, we estimate the direct causal effect of attention to the final prediction, and then maximize such effect to guide attention attending to more meaningful neighbors. Our method can serve as a plug-and-play module for any canonical attention-based GNNs in an end-to-end fashion. Extensive experiments on a wide range of benchmark datasets illustrated that, by directly supervising attention functions, the model is able to converge faster with a clearer decision boundary, and thus yields better performances.

Compositional generalization--understanding unseen combinations of seen primitives--is an essential reasoning capability in human intelligence. The AI community mainly studies this capability by fine-tuning neural networks on lots of training samples, while it is still unclear whether and how in-context learning--the prevailing few-shot paradigm based on large language models--exhibits compositional generalization. In this paper, we present CoFe, a test suite to investigate in-context compositional generalization. We find that the compositional generalization performance can be easily affected by the selection of in-context examples, thus raising the research question what the key factors are to make good in-context examples for compositional generalization. We study three potential factors: similarity, diversity and complexity. Our systematic experiments indicate that in-context examples should be structurally similar to the test case, diverse from each other, and individually simple. Furthermore, two strong limitations are observed: in-context compositional generalization on fictional words is much weaker than that on commonly used ones; it is still critical that the in-context examples should cover required linguistic structures, even though the backbone model has been pre-trained on large corpus. We hope our analysis would facilitate the understanding and utilization of in-context learning paradigm., Comment: ACL 2023 main conference, long paper

Non-convex optimization plays a key role in a growing number of machine learning applications. This motivates the identification of specialized structure that enables sharper theoretical analysis. One such identified structure is quasar-convexity, a non-convex generalization of convexity that subsumes convex functions. Existing algorithms for minimizing quasar-convex functions in the stochastic setting have either high complexity or slow convergence, which prompts us to derive a new class of stochastic methods for optimizing smooth quasar-convex functions. We demonstrate that our algorithms have fast convergence and outperform existing algorithms on several examples, including the classical problem of learning linear dynamical systems. We also present a unified analysis of our newly proposed algorithms and a previously studied deterministic algorithm., Comment: Accepted at the main conference of ICML 2023. 30 pages