An Liu

Yang Jin, Kun Xu, Kun Xu et al. · pku

Recently, the remarkable advance of the Large Language Model (LLM) has inspired researchers to transfer its extraordinary reasoning capability to both vision and language data. However, the prevailing approaches primarily regard the visual input as a prompt and focus exclusively on optimizing the text generation process conditioned upon vision content by a frozen LLM. Such an inequitable treatment of vision and language heavily constrains the model's potential. In this paper, we break through this limitation by representing both vision and language in a unified form. Specifically, we introduce a well-designed visual tokenizer to translate the non-linguistic image into a sequence of discrete tokens like a foreign language that LLM can read. The resulting visual tokens encompass high-level semantics worthy of a word and also support dynamic sequence length varying from the image. Coped with this tokenizer, the presented foundation model called LaVIT can handle both image and text indiscriminately under the same generative learning paradigm. This unification empowers LaVIT to serve as an impressive generalist interface to understand and generate multi-modal content simultaneously. Extensive experiments further showcase that it outperforms the existing models by a large margin on massive vision-language tasks. Our code and models are available at https://github.com/jy0205/LaVIT.

Kexuan Wang, An Liu, Baishuo Lin

Deep Actor-Critic algorithms, which combine Actor-Critic with deep neural network (DNN), have been among the most prevalent reinforcement learning algorithms for decision-making problems in simulated environments. However, the existing deep Actor-Critic algorithms are still not mature to solve realistic problems with non-convex stochastic constraints and high cost to interact with the environment. In this paper, we propose a single-loop deep Actor-Critic (SLDAC) algorithmic framework for general constrained reinforcement learning (CRL) problems. In the actor step, the constrained stochastic successive convex approximation (CSSCA) method is applied to handle the non-convex stochastic objective and constraints. In the critic step, the critic DNNs are only updated once or a few finite times for each iteration, which simplifies the algorithm to a single-loop framework (the existing works require a sufficient number of updates for the critic step to ensure a good enough convergence of the inner loop for each iteration). Moreover, the variance of the policy gradient estimation is reduced by reusing observations from the old policy. The single-loop design and the observation reuse effectively reduce the agent-environment interaction cost and computational complexity. In spite of the biased policy gradient estimation incurred by the single-loop design and observation reuse, we prove that the SLDAC with a feasible initial point can converge to a Karush-Kuhn-Tuker (KKT) point of the original problem almost surely. Simulations show that the SLDAC algorithm can achieve superior performance with much lower interaction cost.

Zhili Cheng, Zhitong Wang, Jinyi Hu et al. · tsinghua

Despite advancements in Large Language Models (LLMs) and Large Multimodal Models (LMMs), their integration into language-grounded, human-like embodied agents remains incomplete, hindering complex real-life task performance in physical environments. Existing integrations often feature limited open sourcing, challenging collective progress in this field. We introduce LEGENT, an open, scalable platform for developing embodied agents using LLMs and LMMs. LEGENT offers a dual approach: a rich, interactive 3D environment with communicable and actionable agents, paired with a user-friendly interface, and a sophisticated data generation pipeline utilizing advanced algorithms to exploit supervision from simulated worlds at scale. In our experiments, an embryonic vision-language-action model trained on LEGENT-generated data surpasses GPT-4V in embodied tasks, showcasing promising generalization capabilities.

Luyuan Zhang, Xidong Mu, An Liu et al.

Pinching antenna systems (PASS) have been proposed as a revolutionary flexible antenna technology which facilitates line-of-sight links via numerous low-cost pinching antennas with adjustable activation positions over waveguides. This letter proposes a two-timescale joint transmit and pinching beamforming design for the maximization of sum rate of a PASS-based downlink multi-user multiple input single output system. A primal dual decomposition method is developed to decouple the two-timescale problem into two sub-problems: 1) A Karush-Kuhn-Tucker-guided dual learning-based approach is proposed to solve the short-term transmit beamforming design sub-problem; 2) The long-term pinching beamforming design sub-problem is tackled by adopting a stochastic successive convex approximation method. Simulation results demonstrate that the proposed two-timescale algorithm achieves a significant performance gain compared to other baselines.

Zonghan Yang, An Liu, Zijun Liu et al. · tsinghua

The rapid progress of foundation models has led to the prosperity of autonomous agents, which leverage the universal capabilities of foundation models to conduct reasoning, decision-making, and environmental interaction. However, the efficacy of agents remains limited when operating in intricate, realistic environments. In this work, we introduce the principles of $\mathbf{U}$nified $\mathbf{A}$lignment for $\mathbf{A}$gents ($\mathbf{UA}^2$), which advocate for the simultaneous alignment of agents with human intentions, environmental dynamics, and self-constraints such as the limitation of monetary budgets. From the perspective of $\mathbf{UA}^2$, we review the current agent research and highlight the neglected factors in existing agent benchmarks and method candidates. We also conduct proof-of-concept studies by introducing realistic features to WebShop, including user profiles to demonstrate intentions, personalized reranking for complex environmental dynamics, and runtime cost statistics to reflect self-constraints. We then follow the principles of $\mathbf{UA}^2$ to propose an initial design of our agent, and benchmark its performance with several candidate baselines in the retrofitted WebShop. The extensive experimental results further prove the importance of the principles of $\mathbf{UA}^2$. Our research sheds light on the next steps of autonomous agent research with improved general problem-solving abilities.

An Liu, Zonghan Yang, Zhenhe Zhang et al. · tsinghua

While Large language models (LLMs) have demonstrated considerable capabilities across various natural language tasks, they often fall short of the performance achieved by domain-specific state-of-the-art models. One potential approach to enhance domain-specific capabilities of LLMs involves fine-tuning them using corresponding datasets. However, this method can be both resource and time-intensive, and not applicable to closed-source commercial LLMs. In this paper, we propose Preference Adaptation for Enhancing Domain-specific Abilities of LLMs (PANDA), a method designed to augment the domain-specific capabilities of LLMs by leveraging insights from the response preference of expert models without requiring fine-tuning. Our experimental results reveal that PANDA significantly enhances the domain-specific ability of LLMs on text classification and interactive decision tasks. Moreover, LLM with PANDA even outperforms the expert model that being learned on 4 tasks of ScienceWorld. This finding highlights the potential of exploring tuning-free approaches to achieve weak-to-strong generalization.

Luyuan Zhang, An Liu, Kexuan Wang

In the forthcoming 6G era, extend reality (XR) has been regarded as an emerging application for ultra-reliable and low latency communications (URLLC) with new traffic characteristics and more stringent requirements. In addition to the quasi-periodical traffic in XR, burst traffic with both large frame size and random arrivals in some real world low latency communication scenarios has become the leading cause of network congestion or even collapse, and there still lacks an efficient algorithm for the resource scheduling problem under burst traffic with hard latency constraints. We propose a novel hybrid reinforcement learning framework for resource scheduling with hard latency constraints (HRL-RSHLC), which reuses polices from both old policies learned under other similar environments and domain-knowledge-based (DK) policies constructed using expert knowledge to improve the performance. The joint optimization of the policy reuse probabilities and new policy is formulated as an Markov Decision Problem (MDP), which maximizes the hard-latency constrained effective throughput (HLC-ET) of users. We prove that the proposed HRL-RSHLC can converge to KKT points with an arbitrary initial point. Simulations show that HRL-RSHLC can achieve superior performance with faster convergence speed compared to baseline algorithms.

Daojun Chen, Xi Wang, Shenyuan Ren et al.

While Large Language Models have significantly advanced Text2SQL generation, a critical semantic gap persists where syntactically valid queries often misinterpret user intent. To mitigate this challenge, we propose GBV-SQL, a novel multi-agent framework that introduces Guided Generation with SQL2Text Back-translation Validation. This mechanism uses a specialized agent to translate the generated SQL back into natural language, which verifies its logical alignment with the original question. Critically, our investigation reveals that current evaluation is undermined by a systemic issue: the poor quality of the benchmarks themselves. We introduce a formal typology for "Gold Errors", which are pervasive flaws in the ground-truth data, and demonstrate how they obscure true model performance. On the challenging BIRD benchmark, GBV-SQL achieves 63.23% execution accuracy, a 5.8% absolute improvement. After removing flawed examples, GBV-SQL achieves 96.5% (dev) and 97.6% (test) execution accuracy on the Spider benchmark. Our work offers both a robust framework for semantic validation and a critical perspective on benchmark integrity, highlighting the need for more rigorous dataset curation.

Kexuan Wang, An Liu

In XR downlink transmission, energy-efficient power scheduling (EEPS) is essential for conserving power resource while delivering large data packets within hard-latency constraints. Traditional constrained reinforcement learning (CRL) algorithms show promise in EEPS but still struggle with non-convex stochastic constraints, non-stationary data traffic, and sparse delayed packet dropout feedback (rewards) in XR. To overcome these challenges, this paper models the EEPS in XR as a dynamic parameter-constrained Markov decision process (DP-CMDP) with a varying transition function linked to the non-stationary data traffic and solves it by a proposed context-aware constrained reinforcement learning (CACRL) algorithm, which consists of a context inference (CI) module and a CRL module. The CI module trains an encoder and multiple potential networks to characterize the current transition function and reshape the packet dropout rewards according to the context, transforming the original DP-CMDP into a general CMDP with immediate dense rewards. The CRL module employs a policy network to make EEPS decisions under this CMDP and optimizes the policy using a constrained stochastic successive convex approximation (CSSCA) method, which is better suited for non-convex stochastic constraints. Finally, theoretical analyses provide deep insights into the CADAC algorithm, while extensive simulations demonstrate that it outperforms advanced baselines in both power conservation and satisfying packet dropout constraints.

Chang Tian, An Liu, Guang Huang et al.

We propose a successive convex approximation based off-policy optimization (SCAOPO) algorithm to solve the general constrained reinforcement learning problem, which is formulated as a constrained Markov decision process (CMDP) in the context of average cost. The SCAOPO is based on solving a sequence of convex objective/feasibility optimization problems obtained by replacing the objective and constraint functions in the original problems with convex surrogate functions. At each iteration, the convex surrogate problem can be efficiently solved by Lagrange dual method even the policy is parameterized by a high-dimensional function. Moreover, the SCAOPO enables to reuse old experiences from previous updates, thereby significantly reducing the implementation cost when deployed in the real-world engineering systems that need to online learn the environment. In spite of the time-varying state distribution and the stochastic bias incurred by the off-policy learning, the SCAOPO with a feasible initial point can still provably converge to a Karush-Kuhn-Tucker (KKT) point of the original problem almost surely.

An Liu, Rui Yang, Tony Q. S. Quek et al.

We consider a two-stage stochastic optimization problem, in which a long-term optimization variable is coupled with a set of short-term optimization variables in both objective and constraint functions. Despite that two-stage stochastic optimization plays a critical role in various engineering and scientific applications, there still lack efficient algorithms, especially when the long-term and short-term variables are coupled in the constraints. To overcome the challenge caused by tightly coupled stochastic constraints, we first establish a two-stage primal-dual decomposition (PDD) method to decompose the two-stage problem into a long-term problem and a family of short-term subproblems. Then we propose a PDD-based stochastic successive convex approximation (PDD-SSCA) algorithmic framework to find KKT solutions for two-stage stochastic optimization problems. At each iteration, PDD-SSCA first runs a short-term sub-algorithm to find stationary points of the short-term subproblems associated with a mini-batch of the state samples. Then it constructs a convex surrogate for the long-term problem based on the deep unrolling of the short-term sub-algorithm and the back propagation method. Finally, the optimal solution of the convex surrogate problem is solved to generate the next iterate. We establish the almost sure convergence of PDD-SSCA and customize the algorithmic framework to solve two important application problems. Simulations show that PDD-SSCA can achieve superior performance over existing solutions.

Jisheng Dai, An Liu, Hing Cheung So

This study addresses the problem of discrete signal reconstruction from the perspective of sparse Bayesian learning (SBL). Generally, it is intractable to perform the Bayesian inference with the ideal discretization prior under the SBL framework. To overcome this challenge, we introduce a novel discretization enforcing prior to exploit the knowledge of the discrete nature of the signal-of-interest. By integrating the discretization enforcing prior into the SBL framework and applying the variational Bayesian inference (VBI) methodology, we devise an alternating optimization algorithm to jointly characterize the finite-alphabet feature and reconstruct the unknown signal. When the measurement matrix is i.i.d. Gaussian per component, we further embed the generalized approximate message passing (GAMP) into the VBI-based method, so as to directly adopt the ideal prior and significantly reduce the computational burden. Simulation results demonstrate substantial performance improvement of the two proposed methods over existing schemes. Moreover, the GAMP-based variant outperforms the VBI-based method with i.i.d. Gaussian measurement matrices but it fails to work for non i.i.d. Gaussian matrices.

An Liu, Vincent Lau

We propose a cache-enabled opportunistic cooperative MIMO (CoMP) framework for wireless video streaming. By caching a portion of the video files at the relays (RS) using a novel MDS-coded random cache scheme, the base station (BS) and RSs opportunistically employ CoMP to achieve spatial multiplexing gain without expensive payload backhaul. We study a two timescale joint optimization of power and cache control to support real-time video streaming. The cache control is to create more CoMP opportunities and is adaptive to the long-term popularity of the video files. The power control is to guarantee the QoS requirements and is adaptive to the channel state information (CSI), the cache state at the RS and the queue state information (QSI) at the users. The joint problem is decomposed into an inner power control problem and an outer cache control problem. We first derive a closed-form power control policy from an approximated Bellman equation. Based on this, we transform the outer problem into a convex stochastic optimization problem and propose a stochastic subgradient algorithm to solve it. Finally, the proposed solution is shown to be asymptotically optimal for high SNR and small timeslot duration. Its superior performance over various baselines is verified by simulations.