Zhan Shu

Bang Li, Donghao Luo, Yujie Liang et al. · tencent-ai

Oracle bone inscriptions(OBI) is the earliest developed writing system in China, bearing invaluable written exemplifications of early Shang history and paleography. However, the task of deciphering OBI, in the current climate of the scholarship, can prove extremely challenging. Out of the 4,500 oracle bone characters excavated, only a third have been successfully identified. Therefore, leveraging the advantages of advanced AI technology to assist in the decipherment of OBI is a highly essential research topic. However, fully utilizing AI's capabilities in these matters is reliant on having a comprehensive and high-quality annotated OBI dataset at hand whereas most existing datasets are only annotated in just a single or a few dimensions, limiting the value of their potential application. For instance, the Oracle-MNIST dataset only offers 30k images classified into 10 categories. Therefore, this paper proposes an Oracle Bone Inscriptions Multi-modal Dataset(OBIMD), which includes annotation information for 10,077 pieces of oracle bones. Each piece has two modalities: pixel-level aligned rubbings and facsimiles. The dataset annotates the detection boxes, character categories, transcriptions, corresponding inscription groups, and reading sequences in the groups of each oracle bone character, providing a comprehensive and high-quality level of annotations. This dataset can be used for a variety of AI-related research tasks relevant to the field of OBI, such as OBI Character Detection and Recognition, Rubbing Denoising, Character Matching, Character Generation, Reading Sequence Prediction, Missing Characters Completion task and so on. We believe that the creation and publication of a dataset like this will help significantly advance the application of AI algorithms in the field of OBI research.

Zhehua Zhou, Jiayang Song, Kunpeng Yao et al.

Motivated by the substantial achievements observed in Large Language Models (LLMs) in the field of natural language processing, recent research has commenced investigations into the application of LLMs for complex, long-horizon sequential task planning challenges in robotics. LLMs are advantageous in offering the potential to enhance the generalizability as task-agnostic planners and facilitate flexible interaction between human instructors and planning systems. However, task plans generated by LLMs often lack feasibility and correctness. To address this challenge, we introduce ISR-LLM, a novel framework that improves LLM-based planning through an iterative self-refinement process. The framework operates through three sequential steps: preprocessing, planning, and iterative self-refinement. During preprocessing, an LLM translator is employed to convert natural language input into a Planning Domain Definition Language (PDDL) formulation. In the planning phase, an LLM planner formulates an initial plan, which is then assessed and refined in the iterative self-refinement step by using a validator. We examine the performance of ISR-LLM across three distinct planning domains. The results show that ISR-LLM is able to achieve markedly higher success rates in task accomplishments compared to state-of-the-art LLM-based planners. Moreover, it also preserves the broad applicability and generalizability of working with natural language instructions.

Jiayang Song, Zhehua Zhou, Jiawei Liu et al.

Although Deep Reinforcement Learning (DRL) has achieved notable success in numerous robotic applications, designing a high-performing reward function remains a challenging task that often requires substantial manual input. Recently, Large Language Models (LLMs) have been extensively adopted to address tasks demanding in-depth common-sense knowledge, such as reasoning and planning. Recognizing that reward function design is also inherently linked to such knowledge, LLM offers a promising potential in this context. Motivated by this, we propose in this work a novel LLM framework with a self-refinement mechanism for automated reward function design. The framework commences with the LLM formulating an initial reward function based on natural language inputs. Then, the performance of the reward function is assessed, and the results are presented back to the LLM for guiding its self-refinement process. We examine the performance of our proposed framework through a variety of continuous robotic control tasks across three diverse robotic systems. The results indicate that our LLM-designed reward functions are able to rival or even surpass manually designed reward functions, highlighting the efficacy and applicability of our approach.

Gengluo Li, Shangpin Peng, Xingyu Wan et al.

Vision Large Language Models (VLLMs) have achieved remarkable success in modern text-rich visual understanding. However, their perceptual robustness in the face of the continuous morphological evolution of historical writing systems remains largely unexplored. Existing ancient text datasets typically focus on isolated historical periods, failing to capture the systematic visual distribution shifts spanning thousands of years. To bridge this gap and empower Digital Humanities, we introduce Chronicles-OCR, the first comprehensive benchmark specifically designed to evaluate the cross-temporal visual perception capabilities of VLLMs across the complete evolutionary trajectory of Chinese characters, known as the Seven Chinese Scripts. Curated in collaboration with top-tier institutional domain experts, the dataset comprises 2,800 strictly balanced images encompassing highly diverse physical media, ranging from tortoise shells to paper-based calligraphy. To accommodate the drastic morphological and topological variations across different historical stages, we propose a novel Stage-Adaptive Annotation Paradigm. Based on this, Chronicles-OCR formulates four rigorous quantitative tasks: cross-period character spotting, fine-grained archaic character recognition via visual referring, ancient text parsing, and script classification. By isolating visual perception from semantic reasoning, Chronicles-OCR provides an authoritative platform to expose the limitations of current VLLMs, paving the way for robust, evolution-aware historical text perception. Chronicles-OCR is publicly available at https://github.com/VirtualLUOUCAS/Chronicles-OCR.

Wei Zhang, Dai Li, Chen Liang et al.

Effective user representations are pivotal in personalized advertising. However, stringent constraints on training throughput, serving latency, and memory, often limit the complexity and input feature set of online ads ranking models. This challenge is magnified in extensive systems like Meta's, which encompass hundreds of models with diverse specifications, rendering the tailoring of user representation learning for each model impractical. To address these challenges, we present Scaling User Modeling (SUM), a framework widely deployed in Meta's ads ranking system, designed to facilitate efficient and scalable sharing of online user representation across hundreds of ads models. SUM leverages a few designated upstream user models to synthesize user embeddings from massive amounts of user features with advanced modeling techniques. These embeddings then serve as inputs to downstream online ads ranking models, promoting efficient representation sharing. To adapt to the dynamic nature of user features and ensure embedding freshness, we designed SUM Online Asynchronous Platform (SOAP), a latency free online serving system complemented with model freshness and embedding stabilization, which enables frequent user model updates and online inference of user embeddings upon each user request. We share our hands-on deployment experiences for the SUM framework and validate its superiority through comprehensive experiments. To date, SUM has been launched to hundreds of ads ranking models in Meta, processing hundreds of billions of user requests daily, yielding significant online metric gains and improved infrastructure efficiency.

Xin Gong, Jintao Peng, Dong Yang et al.

This paper studies the leader-following consensuses of uncertain and nonlinear multi-agent systems against composite attacks (CAs), including Denial of Service (DoS) attacks and actuation attacks (AAs). A double-layer control framework is formulated, where a digital twin layer (TL) is added beside the traditional cyber-physical layer (CPL), inspired by the recent Digital Twin technology. Consequently, the resilient control task against CAs can be divided into two parts: One is distributed estimation against DoS attacks on the TL and the other is resilient decentralized tracking control against actuation attacks on the CPL. %The data-driven scheme is used to deal with both model non-linearity and model uncertainty, in which only the input and output data of the system are employed throughout the whole control process. First, a distributed observer based on switching estimation law against DoS is designed on TL. Second, a distributed model free adaptive control (DMFAC) protocol based on attack compensation against AAs is designed on CPL. Moreover, the uniformly ultimately bounded convergence of consensus error of the proposed double-layer DMFAC algorithm is strictly proved. Finally, the simulation verifies the effectiveness of the resilient double-layer control scheme.

Matin Macktoobian, Zhan Shu, Qing Zhao

Faults occurring in ad-hoc robot networks may fatally perturb their topologies leading to disconnection of subsets of those networks. Optimal topology synthesis is generally resource-intensive and time-consuming to be done in real time for large ad-hoc robot networks. One should only perform topology re-computations if the probability of topology recoverability after the occurrence of any fault surpasses that of its irrecoverability. We formulate this problem as a binary classification problem. Then, we develop a two-pathway data-driven model based on Bayesian Gaussian mixture models that predicts the solution to a typical problem by two different pre-fault and post-fault prediction pathways. The results, obtained by the integration of the predictions of those pathways, clearly indicate the success of our model in solving the topology (ir)recoverability prediction problem compared to the best of current strategies found in the literature.

Syed Eqbal Alam, Zhan Shu

We develop algorithms for collaborative control of AI agents and critics in a multi-actor, multi-critic federated multi-agent system. Each AI agent and critic has access to classical machine learning or generative AI foundation models. The AI agents and critics collaborate with a central server to complete multimodal tasks such as fault detection, severity, and cause analysis in a network telemetry system, text-to-image generation, video generation, healthcare diagnostics from medical images and patient records, etcetera. The AI agents complete their tasks and send them to AI critics for evaluation. The critics then send feedback to agents to improve their responses. Collaboratively, they minimize the overall cost to the system with no inter-agent or inter-critic communication. AI agents and critics keep their cost functions or derivatives of cost functions private. Using multi-time scale stochastic approximation techniques, we provide convergence guarantees on the time-average active states of AI agents and critics. The communication overhead is a little on the system, of the order of $\mathcal{O}(m)$, for $m$ modalities and is independent of the number of AI agents and critics. Finally, we present an example of fault detection, severity, and cause analysis in network telemetry and thorough evaluation to check the algorithm's efficacy.

Zhehua Zhou, Xuan Xie, Jiayang Song et al.

Safe Reinforcement Learning (SRL) aims to realize a safe learning process for Deep Reinforcement Learning (DRL) algorithms by incorporating safety constraints. However, the efficacy of SRL approaches often relies on accurate function approximations, which are notably challenging to achieve in the early learning stages due to data insufficiency. To address this issue, we introduce in this work a novel Generalizable Safety enhancer (GenSafe) that is able to overcome the challenge of data insufficiency and enhance the performance of SRL approaches. Leveraging model order reduction techniques, we first propose an innovative method to construct a Reduced Order Markov Decision Process (ROMDP) as a low-dimensional approximator of the original safety constraints. Then, by solving the reformulated ROMDP-based constraints, GenSafe refines the actions of the agent to increase the possibility of constraint satisfaction. Essentially, GenSafe acts as an additional safety layer for SRL algorithms. We evaluate GenSafe on multiple SRL approaches and benchmark problems. The results demonstrate its capability to improve safety performance, especially in the early learning phases, while maintaining satisfactory task performance. Our proposed GenSafe not only offers a novel measure to augment existing SRL methods but also shows broad compatibility with various SRL algorithms, making it applicable to a wide range of systems and SRL problems.

Matin Macktoobian, Zhan Shu, Qing Zhao

In this paper, we synthesize a data-driven method to predict the optimal topology of an ad-hoc robot network. This problem is technically a multi-task classification problem. However, we divide it into a class of multi-class classification problems that can be more efficiently solved. For this purpose, we first compose an algorithm to create ground-truth optimal topologies associated with various configurations of a robot network. This algorithm incorporates a complex collection of optimality criteria that our learning model successfully manages to learn. This model is an stacked ensemble whose output is the topology prediction for a particular robot. Each stacked ensemble instance constitutes three low-level estimators whose outputs will be aggregated by a high-level boosting blender. Applying our model to a network of 10 robots displays over 80% accuracy in the prediction of optimal topologies corresponding to various configurations of the cited network.

Arash Mohammadhasani, Hamed Mehrivash, Alan Lynch et al.

In this paper, we develop a safe decision-making method for self-driving cars in a multi-lane, single-agent setting. The proposed approach utilizes deep reinforcement learning (RL) to achieve a high-level policy for safe tactical decision-making. We address two major challenges that arise solely in autonomous navigation. First, the proposed algorithm ensures that collisions never happen, and therefore accelerate the learning process. Second, the proposed algorithm takes into account the unobservable states in the environment. These states appear mainly due to the unpredictable behavior of other agents, such as cars, and pedestrians, and make the Markov Decision Process (MDP) problematic when dealing with autonomous navigation. Simulations from a well-known self-driving car simulator demonstrate the applicability of the proposed method

Jeyamohan Neera, Xiaomin Chen, Nauman Aslam et al.

Recommendation systems rely heavily on users behavioural and preferential data (e.g. ratings, likes) to produce accurate recommendations. However, users experience privacy concerns due to unethical data aggregation and analytical practices carried out by the Service Providers (SP). Local differential privacy (LDP) based perturbation mechanisms add noise to users data at user side before sending it to the SP. The SP then uses the perturbed data to perform recommendations. Although LDP protects the privacy of users from SP, it causes a substantial decline in predictive accuracy. To address this issue, we propose an LDP-based Matrix Factorization (MF) with a Gaussian Mixture Model (MoG). The LDP perturbation mechanism, Bounded Laplace (BLP), regulates the effect of noise by confining the perturbed ratings to a predetermined domain. We derive a sufficient condition of the scale parameter for BLP to satisfy $ε$ LDP. At the SP, The MoG model estimates the noise added to perturbed ratings and the MF algorithm predicts missing ratings. Our proposed LDP based recommendation system improves the recommendation accuracy without violating LDP principles. The empirical evaluations carried out on three real world datasets, i.e., Movielens, Libimseti and Jester, demonstrate that our method offers a substantial increase in predictive accuracy under strong privacy guarantee.

Jing Lai, Junlin Xiong, Zhan Shu

This paper investigates the optimal control problem for a class of discrete-time stochastic systems subject to additive and multiplicative noises. A stochastic Lyapunov equation and a stochastic algebra Riccati equation are established for the existence of the optimal admissible control policy. A model-free reinforcement learning algorithm is proposed to learn the optimal admissible control policy using the data of the system states and inputs without requiring any knowledge of the system matrices. It is proven that the learning algorithm converges to the optimal admissible control policy. The implementation of the model-free algorithm is based on batch least squares and numerical average. The proposed algorithm is illustrated through a numerical example, which shows our algorithm outperforms other policy iteration algorithms.