United States

Constrained Markov decision processes (CMDPs), in which the agent optimizes expected payoffs while keeping the expected cost below a given threshold, are the leading framework for safe sequential decision making under stochastic uncertainty. Among algorithms for planning and learning in CMDPs, methods based on Monte Carlo tree search (MCTS) have particular importance due to their efficiency and extendibility to more complex frameworks (such as partially observable settings and games). However, current MCTS-based methods for CMDPs either struggle with finding safe (i.e., constraint-satisfying) policies, or are too conservative and do not find valuable policies. We introduce Threshold UCT (T-UCT), an online MCTS-based algorithm for CMDP planning. Unlike previous MCTS-based CMDP planners, T-UCT explicitly estimates Pareto curves of cost-utility trade-offs throughout the search tree, using these together with a novel action selection and threshold update rule to seek safe and valuable policies. Our experiments demonstrate that our approach significantly outperforms state-of-the-art methods from the literature.

AAAI 2025

Threshold UCT: Cost-Constrained Monte Carlo Tree Search with Pareto Curves

planning under uncertainty

The purpose of the AAAI conference is to promote research in artificial intelligence (AI) and scientific exchange among AI researchers, practitioners, scientists, and engineers in affiliated disciplines. 

AAAI 2021

technical paper

We are pleased to announce the Thirty-Ninth AAAI Conference on Artificial Intelligence (AAAI-25), which will be held in Philadelphia, Pennsylvania at the Pennsylvania Convention Center from February 25 to March 4, 2025.

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-25 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and a range of other activities to be announced.

### [Invited Speakers](https://aaai.org/conference/aaai/aaai-25/aaai-25-invited-speakers/)

Register [here](https://aaai.org/conference/aaai/aaai-25/registration/)

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-25 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and a range of other activities to be announced.



We focus on the medication recommendation problem aiming to recommend accurate medications for a patient’s current visit. Most existing methods for this problem utilize the patient’s current health status, medications prescribed at her past visits, and an Electronic Health Records (EHR) graph which represents whether medications have been co-prescribed. However, we point out their two limitations: (1) they have difficulty in utilizing only the medications which have been prescribed in health status similar to the patient’s current health status, regardless of whether they are prescribed at her past visits or at other patients’ visits; (2) for two medications that have ever been co-prescribed, their EHR graph does not consider the degree to which one medication is prescribed together when the other is prescribed. To address these two limitations, we propose a novel medication recommendation framework, named HI-DR (pronounced as ‘Hi Doctor’), composed of following two core ideas: (Idea 1) Health status-aware attentIon; (Idea 2) an electronic health recorDs gRaph+. Extensive experiments on real-world datasets demonstrate the significant superiority of HI-DR (up to 18.69% higher accuracy than the best competitor) and the effectiveness of two core ideas in HI-DR.

HI-DR: Exploiting Health Status-Aware Attention and an EHR Graph+ for Effective Medication Recommendation

Coalition formation concerns autonomous agents that strategically interact to form self-organized coalitions. When agents lack initial sufficient information to evaluate their preferences before interacting with others, they learn them online through repeated feedback while iteratively forming coalitions. In this work, we introduce online learning in coalition formation from a non-cooperative perspective, studying the impact of collective data utilization where selfish agents aim to accelerate their learning by leveraging a shared data platform. Thus, the efficiency and dynamics of the learning process are affected by each agent's local feedbacks, motivating us to explore the tension between semi-bandit and bandit feedback, which differ in the granularity of utility information observed by each agent. Under our non-cooperative viewpoint, we evaluate the system by means of Nash stability, where no agent can improve her utility by unilaterally deviating. Our main result is a sample-efficient algorithm for selfish agents that aims to minimize their Nash regret under both semi-bandit and bandit feedback, implying approximately Nash stable outcomes. Under both feedback settings, our algorithm enjoys Nash regret and sample complexity bounds that are optimal up to logarithmic factors.

Online Learning of Coalition Structures by Selfish Agents

The MaxSAT problem is an optimization version of the satisfiability problem (SAT). A tight lower bound (LB) on the number of the falsified soft clauses in a MaxSAT solution is crucial for the efficiency of Branch-and-Bound (BnB) MaxSAT solvers. To compute LB, modern BnB solvers detect disjoint inconsistent subsets of soft clauses, called cores, using unit propagation. A notable feature in these solvers is that soft clauses belonging to already detected cores cannot be reused to detect additional cores, limiting the number of cores that can be detected. In this paper, we propose an unlocking mechanism that allows the reuse of soft clauses in already detected cores while ensuring the soundness of LB. Experimental results show that this unlocking mechanism consistently improves the performance of a state-of-the-art BnB solver.

Improving the Lower Bound in Branch-and-Bound Algorithms for MaxSAT

There are two manifestations of classification fairness. One is the preference for head classes with more instances due to the long-tail (LT) distribution of training data. The other is the clever Hans (CH) effect, where non-discriminative features are mistakenly used for classification. In this paper, we find that using category-agnostic zero-valued data can simultaneously reveal both types of unfairness. Based on this, we propose a zero uniformity training (ZUT) framework to optimize classification fairness. The ZUT framework inputs category-agnostic zero-valued data into the model in parallel and uses zero uniformity loss (ZUL) to optimize classification fairness. The ZUL loss mitigates bias towards specific classes by unifying the classification features corresponding to zero-valued data. The ZUT framework is compatible with various classification-based tasks. Experiments show that the ZUT framework can improve the performance of multiple state-of-the-art methods in image classification, person re-identification, and semantic segmentation.

Fair Training with Zero Inputs

Accurately predicting the trajectory of vehicles is critically important for ensuring safety and reliability in autonomous driving. Although considerable research efforts have been made recently, the inherent trajectory uncertainty caused by various factors including the dynamic driving intends and the diverse driving scenarios still poses significant challenges to accurate trajectory prediction. To address this issue, we propose C2F-TP, a coarse-to-fine denoising framework for uncertainty-aware vehicle trajectory prediction. C2F-TP features an innovative two-stage coarse-to-fine prediction process. Specifically, in the first stage, we propose a spatial-temporal interaction module to capture the inter-vehicle interactions and learn a multimodal trajectory distribution, from which a certain number of noisy trajectories are sampled. Next, in the trajectory refinement stage, we design a conditional denoising model to reduce the uncertainties of the sampled trajectories through a step-wise denoising operation. Extensive experiments are conducted on two real datasets NGSIM and highD that are widely adopted in trajectory prediction. The result demonstrates the effectiveness of our proposal. The code can be found at https://anonymous.4open.science/r/C2F-TP-1459.

C2F-TP: A Coarse-to-Fine Denoising Framework for Uncertainty-Aware Trajectory Prediction

Markov decision processes (MDP) are a well-established model for sequential decision-making in the presence of probabilities.
	In *robust* MDP (RMDP), every action is associated with an *uncertainty set* of probability distributions, modelling that transition probabilities are not known precisely.
	Based on the known theoretical connection to stochastic games, we provide a framework for solving RMDPs that is generic, reliable, and efficient.
	It is *generic* both with respect to the model, allowing for a wide range of uncertainty sets, including but not limited to intervals, $L^1$- or $L^2$-balls, and polytopes; and with respect to the objective, including long-run average reward, undiscounted total reward, and stochastic shortest path.
	It is *reliable*, as our approach not only converges in the limit, but provides precision guarantees at any time during the computation.
	It is *efficient* because -- in contrast to state-of-the-art approaches -- it avoids explicitly constructing the underlying stochastic game.
	Consequently, our prototype implementation outperforms existing tools by several orders of magnitude and can solve RMDPs with a million states in under a minute.

Solving Robust Markov Decision Processes: Generic, Reliable, Efficient

Multi-objective evolutionary algorithms (MOEAs) are widely used for searching optimal solutions in complex multi-component applications. Traditional MOEAs for multi-component deep learning (MCDL) systems face challenges in enhancing the search efficiency while maintaining the diversity. To combat these, this paper proposes $\mu$MOEA, the first LLM-empowered adaptive evolutionary search methodology to detect safety violations in MCDL systems. Inspired by the context-understanding ability of Large Language Models (LLMs), $\mu$MOEA promotes the LLM to comprehend the optimization problem and generate an initial population tailed to evolutionary objectives. Subsequently, it employs adaptive selection and variation techniques to iteratively produce offspring, balancing the evolutionary efficiency and diversity.
During the evolutionary process, to navigate away from the local optima,  $\mu$MOEA integrates the evolutionary experience back into the LLM. This utilization harnesses the LLM's quantitative reasoning prowess to generate differential seeds, breaking away from current optimal solutions. We evaluate the effectiveness of \name in finding safety violations of MCDL systems, and compare its performance with state-of-the-art MOEA methods. Experimental results show that \name can significantly improve the efficiency and diversity of the multi-objective evolutionary search.

An LLM-Empowered Adaptive Evolutionary Algorithm for Multi-Component Deep Learning Systems

Humans naturally rely on floor plans to navigate in unfamiliar environments because they are easy to access, consistently reliable, and rich in geometric detail. However, existing visual navigation settings neglect such informative prior knowledge, limiting the efficiency and accuracy. To eliminate this gap, we introduce a novel navigation task: Floor Plan Visual Navigation (FloNa), the first attempt to incorporate floor plan into the embodied visual navigation. Although the floor plan offers informative knowledge, two additional challenges arise: (1) how to map observed images and the floor plan sketch, given their different modalities, and (2) how to avoid collision with scene furniture that the floor plan does not depict. To address these issues, we propose FloDiff, a novel diffusion policy framework incorporating a localization module to facilitate the matching between the current observation and the floor plan. We further collect 20k navigation episodes from 117 scenes in the iGibson simulator to support the training and evaluation. The extensive experiments show the effectiveness and efficiency of our proposed framework in unfamiliar scenes using floor plan knowledge.

FloNa: Floor Plan Guided Embodied Visual Navigation

In multi-agent reinforcement learning, centralized training with decentralized execution (CTDE) methods typically assume that agents make decisions based on their local observations independently, which may not lead to a correlated joint policy with coordination. Coordination can be explicitly encouraged during training and individual policies can be trained to imitate the correlated joint policy. However, this may lead to an $\textit{asymmetric learning failure}$ due to the observation mismatch between the joint and individual policies. Inspired by the concept of correlated equilibrium, we introduce a $\textit{strategy modification}$ called AgentMixer that allows agents to correlate their policies. AgentMixer combines individual partially observable policies into a joint fully observable policy non-linearly. To enable decentralized execution, we introduce $\textit{Individual-Global-Consistency}$ to guarantee mode consistency during joint training of the centralized and decentralized policies and prove that AgentMixer converges to an $\epsilon$-approximate Correlated Equilibrium. In the Multi-Agent MuJoCo, SMAC-v2, Matrix Game, and Predator-Prey benchmarks, AgentMixer outperforms or matches state-of-the-art methods.

AgentMixer: Multi-Agent Correlated Policy Factorization

Existing semantic segmentation methods face challenges when processing input images degraded by raindrops on the lens or windshield. Unlike other adverse conditions such as fog and nighttime, which degrade visual quality, raindrops not only impair visual appearances but also introduce misleading occlusion, leading to significant performance drops in current models. The novelty of our approach lies in our two-stage, dual teacher-student framework. We tackle the complex problem of raindrop degradation by dividing it into two distinct challenges: degraded visual appearance and raindrop occlusion. These challenges are then addressed individually in two stages, utilizing two pairs of teacher-student networks. This division enables the networks to develop specialized expertise in handling each aspect of raindrop degradation, enabling their collaboration to achieve superior performance. In the first stage, one teacher-student pair focuses on learning to extract information from visual degraded areas. Building on this, the second teacher-student pair focuses specially on the raindrop occlusion. As such, unlike the existing methods, our approach employs a collaborative approach to decompose and address raindrop-induced degradations. In the second stage, we introduce a mask-based recovery technique to identify and rectify areas that likely contain misleading information, thus further refining the predictions. Additionally, this stage encourages both pairs to expand knowledge by swapping their specialized expertise. Our method achieves a performance of 60.3 mIoU on Rainy WCity and 72.8 mIoU on ACDC Rainy, representing an improvement of +4.4 mIoU and +2.3 mIoU over the existing state-of-the-art methods, respectively.

Downloads

Next from AAAI 2025

HI-DR: Exploiting Health Status-Aware Attention and an EHR Graph+ for Effective Medication Recommendation

Similar lecture

Single Player Monte-Carlo Tree Search Based on the Plackett-Luce Model

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES