AAAI 2025 Papers

In complex electromagnetic environments, the identification and differentiation of diverse radio frequency (RF) emitters become particularly crucial. Existing RF fingerprinting methods demonstrate limitations when dealing with numerous unknown emitters, making it challenging for accurate classification and recognition. These limitations hinder the effective handling of specific unknown emitters.To address this issue, we introduce a novel RF fingerprinting method suitable for open-world conditions for the first time. We develop a novel RF fingerprinting model, Roinformer, to extract signal features with positional attention. We then leverage data augmentation strategies such as noise jitter and signal frame rearrangement to construct an effective pre-training model. Moreover, by incorporating instance-level similarity loss and a novel local entropy regularization approach, we significantly enhance the accuracy of known class identification and mitigate the catastrophic forgetting of known signal samples. Experimental results on three temporal signal datasets demonstrate that our method effectively recognizes both the known and unknown classes, outperforming several state-of-the-art methods by a large margin.

With the astonishing ability of different pretrained foundation models (e.g., large language models (LLMs), vision-language models, diffusion models), today’s AI research and development tendency has been revolutionized. In this talk, I will answer two questions: Q1: How can we efficiently train or fine-tune foundation models? Q2: How can we build strong open-world multimodal understanding and generation models with these pretrained foundation models?

Domain adaptation (DA) tackles the issue of distribution shift by learning a model from a source domain that generalizes to a target domain. However, most existing DA methods are designed for scenarios where the source and target domain data lie within the same feature space, which limits their applicability in real-world situations. Recently, heterogeneous DA (HeDA) methods have been introduced to address the challenges posed by heterogeneous feature space between source and target domains. Despite their successes, current HeDA techniques fall short when there is a mismatch in both feature and label spaces. To address this, this paper explores a new DA scenario called open-set HeDA (OSHeDA). In OSHeDA, the model must not only handle heterogeneity in feature space but also identify samples belonging to novel classes. To tackle this challenge, we first develop a novel theoretical framework that constructs learning bounds for prediction error on target domain. Guided by this framework, we propose a new DA method called Representation Learning for OSHeDA (RL-OSHeDA). This method is designed to simultaneously transfer knowledge between heterogeneous data sources and identify novel classes. Experiments across text, image, and clinical data demonstrate the effectiveness of our algorithm.

Existing cross-network node classification methods are mainly proposed for closed-set setting, where the source network and the target network share exactly the same label space. Such a setting is restricted in real-world applications, since the target network might contain additional classes that are not present in the source. In this work, we study a more realistic open-set cross-network node classification (O-CNNC) problem, where the target network contains all the known classes in the source and further contains several target-private classes unseen in the source. Borrowing the concept from open-set domain adaptation, all target-private classes are defined as an additional “unknown” class. To address the challenging O-CNNC problem, we propose an unknown-excluded adversarial graph domain alignment (UAGA) model with a separate-adapt training strategy. Firstly, UAGA roughly separates known classes from unknown class, by training a graph neural network encoder and a neighborhood-aggregation node classifier in an adversarial framework. Then, unknown-excluded adversarial domain alignment is customized to align only target nodes from known classes with the source, while pushing target nodes from unknown class far away from the source, by assigning positive and negative domain adaptation coefficient to known class nodes and unknown class nodes. Extensive experiments on real-world datasets demonstrate significant outperformance of the proposed UAGA over state-of-the-art methods on O-CNNC.

The emergence of unveiling human-like behaviors in Large Language Models (LLMs) has led to a closer connection between NLP and human psychology. However, research on the personalities exhibited by LLMs has largely been confined to limited investigations using individual psychological tests, primarily focusing on a small number of commercially licensed LLMs. This approach overlooks the extensive use and significant advancements observed in open-source LLMs. This work aims to address both the above limitations by conducting an in-depth investigation of a significant body of 12 LLM Agents based on the most representative Open models, through the two most well-known psychological assessment tests, namely Myers-Briggs Type Indicator (MBTI) and Big Five Inventory (BFI). Our approach involves evaluating the intrinsic personality traits of LLM agents and determining the extent to which these agents can mimic human personalities when conditioned by specific personalities and roles. Our findings unveil that (i) each LLM agent showcases distinct human personalities; (ii) personality-conditioned prompting produces varying effects on the agents, with only few successfully mirroring the imposed personality, while most of them being ``closed-minded'' (i.e., they retain their intrinsic traits); and (iii) combining role and personality conditioning can enhance the agents' ability to mimic human personalities. Our work represents a step up in understanding the dense relationship between NLP and human psychology through the lens of LLMs.

Constrained k-submodular maximization is a general framework that captures many discrete optimization problems such as ad allocation, influence maximization, personalized recommendation, and many others. In many of these applications, datasets are large or decisions need to be made in an online manner, which motivates the development of efficient streaming and online algorithms. In this work, we develop single-pass streaming and online algorithms for constrained k-submodular maximization with both monotone and general (possibly non-monotone) objectives subject to cardinality and knapsack constraints. Our algorithms achieve provable constant-factor approximation guarantees which improve upon the state of the art in almost all settings. Moreover, they achieve the fastest known running times and have optimal space usage. We experimentally evaluate our algorithms on instances for ad allocation and other applications, where we observe that our algorithms are practical and scalable, and construct solutions that are comparable in value even to offline greedy algorithms.

Large Language Models (LLMs) demonstrate impressive capabilities in the domain of program synthesis. This level of performance is not, however, universal across all tasks, all LLMs and all prompting styles. There are many areas where one LLM dominates, one prompting style dominates, or where calling a symbolic solver is a better choice than an LLM. A key challenge for the user then, is to identify not only when an LLM is the right choice of solver, and the appropriate LLM to call for a given synthesis task, but also the right way to call it. A non-expert user who makes the wrong choice, incurs a cost both in terms of results (number of tasks solved, and the time it takes to solve them) and financial cost, if using a closed-source language model via a commercial API. We frame this choice as an online learning problem. We use a multi-armed bandit algorithm to select which symbolic solver, or LLM and prompt combination to deploy in order to maximize a given reward function (which may prioritize solving time, number of synthesis tasks solved, or financial cost of solving). We implement an instance of this approach, called \name, and evaluate it on synthesis queries from the literature in ranking function synthesis, from the syntax-guided synthesis competition, and fresh, unseen queries generated from SMT problems. Cyanea solves 37.2 % more queries than the best single solver and achieves results within 4 % of the virtual best solver.

We investigate the online nonsubmodular optimization with delayed feedback in the bandit setting, where the loss function is α-weakly DR-submodular and β-weakly DR-supermodular. Previous work has established an (α,β)-regret bound of O(nd^⅓T^⅔), where n is the dimensionality and d is the maximum delay. However, its regret bound relies on the maximum delay and is thus sensitive to irregular delays. Additionally, it couples the effects of delays and bandit feedback as its bound is the product of the delay term and the O(nT^⅔) regret bound in the bandit setting without delayed feedback. In this paper, we develop two algorithms to address these limitations, respectively. Firstly, we propose a novel method, namely DBGD-NF, which employs the one-point gradient estimator and utilizes all the available estimated gradients in each round to update the decision. It achieves a better O(nd̅^⅓T^⅔) regret bound, which is relevant to the average delay d̅ = 1/T ∑ₜ₌₁ᵀ dₜ <= d. Secondly, we extend DBGD-NF by employing a blocking update mechanism to decouple the joint effect of the delays and bandit feedback, which enjoys an O(n(T^⅔+ √(dT))) regret bound. When d = O(T^⅓), our regret bound matches the O(nT^⅔) bound in the bandit setting without delayed feedback. Compared to our first O(nd̅^⅓T^⅔) bound, it is more advantageous when the maximum delay d = o(d̅^⅔T^⅓). Finally, we conduct experiments on structured sparse learning to demonstrate the superiority of our methods.

In this work, we consider an online robust Markov Decision Process (MDP) where we have the information of finitely many prototypes of the underlying transition kernel. We consider an adaptively updated ambiguity set of the prototypes and propose an algorithm that efficiently identifies the true underlying transition kernel while guaranteeing the performance of the corresponding robust policy. To be more specific, we provide a sublinear regret of the subsequent optimal robust policy. We also provide an early stopping mechanism and a worst-case performance bound of the value function. In numerical experiments, we demonstrate that our method outperforms existing approaches, particularly in the early stage with limited data. This work contributes to robust MDPs by considering possible prior information about the underlying transition probability and online learning, offering both theoretical insights and practical algorithms for improved decision-making under uncertainty.

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.

We study the problem of optimizing a guidance policy capable of dynamically guiding the agents for lifelong Multi-Agent Path Finding based on real-time traffic patterns. Multi-Agent Path Finding (MAPF) focuses on moving multiple agents from their starts to goals without collisions. Its lifelong variant, LMAPF, continuously assigns new goals to agents. In this work, we focus on improving the solution quality of PIBT, a state-of-the-art rule-based LMAPF algorithm, by optimizing a policy to generate adaptive guidance. We design two pipelines to incorporate guidance in PIBT in two different ways. We demonstrate the superiority of the optimized policy over both static guidance and human-designed policies. Additionally, we explore scenarios where task distribution changes over time, a challenging yet common situation in real-world applications that is rarely explored in the literature.

Anti-fraud machine learning systems are perpetually confronted with the significant challenge of concept drift, driven by the continuous and intense evolution of fraudulent techniques. That is, outdated models trained on historical fraudulent behaviors often fall short in addressing the evolving tactics of malicious users over time. The key issue lies in effectively tackling the rapid and significant evolution of fraudsters' behaviors to detect these emerging and unforeseen anomalies. In this paper, we propose a solution by directly accessing real-time data and introducing a lightweight plug-in approach named TRE (Test-time Retrieval-based Representation Enrichment). Considering the similarity among samples, TRE employs a retriever to efficiently identify the top-K most relevant recent samples and implements an aggregation strategy to provide neighboring embeddings to the predictor. It thus adjusts the trained classifiers during the test time, providing them with the information from the latest unlabeled data. Extensive experiments on three large-scale real-world datasets demonstrate the superiority of TRE. By consistently incorporating information from the nearest neighbors, TRE demonstrates high adaptability and surpasses existing methods in performance.

This paper proposes a novel k-medoids approximation algorithm to handle large-scale datasets with reasonable computational time and memory complexity. We develop a local-search algorithm that iteratively improves the medoid selection based on the estimation of the k-medoids objective. A single batch of size m << n provides the estimation, which reduces the required memory size and the number of pairwise dissimilarities computations to O(mn), instead of O(n^2) compared to most k-medoids baselines. We obtain theoretical results highlighting that a batch of size m = O(log(n)) is sufficient to guarantee, with strong probability, the same performance as the original local-search algorithm. Multiple experiments conducted on real datasets of various sizes and dimensions show that our algorithm provides similar performances as state-of-the-art methods such as FasterPAM and BanditPAM++ with a drastically reduced running time.

Generating sketches that accurately reflect the content of reference images presents numerous challenges. Current methods either require paired training data or fail to accommodate a wider range and diversity of sketch styles. While pre-trained diffusion models have shown strong text-based control capabilities for reference-based content sketch generation, state-of-the-art methods still struggle with reference-based sketch generation for given content. The main difficulties lie in (1) balancing content preservation with style enhancement, and (2) representing content image textures at varying levels of abstraction to approximate the reference sketch style. In this paper, we propose a method (Ref2Sketch-SA) that transforms a given content image into a sketch based on a reference sketch. The core strategies include (1) using DDIM Inversion to enhance structural consistency in the sketch generation of content images; (2) injecting noise into the input image during the denoising process to produce a sketch that retains content attributes while aligning with, yet differing in texture from, the reference. Our model demonstrates superior performance across multiple evaluation metrics, including user style preference.

Multi-domain recommendation (MDR) aims to enhance recommendation performance across various domains. However, real-world recommender systems in online platforms often need to handle dozens or even hundreds of domains, far exceeding the capabilities of traditional MDR algorithms, which typically focus on fewer than five domains. Key challenges include a substantial increase in parameter count, high maintenance costs, and intricate knowledge transfer patterns across domains. Furthermore, minor domains often suffer from data sparsity, leading to inadequate training in classical methods. To address these issues, we propose Adaptive REcommendation for All Domains with counterfactual augmentation (AREAD). AREAD employs a hierarchical structure with a limited number of expert networks at several layers, to effectively capture domain knowledge at different granularities. To adaptively capture the knowledge transfer pattern across domains, we generate and iteratively prune a hierarchical expert network selection mask for each domain during training. Additionally, counterfactual assumptions are used to augment data in minor domains, supporting their iterative mask pruning. Our experiments on two public datasets, each encompassing over twenty domains, demonstrate AREAD's effectiveness, especially in data-sparse domains.

Retrieval-augmented generation (RAG) is a promising way to improve large language models (LLMs) for generating more factual, accurate, and up-to-date content. Existing methods either optimize prompts to guide LLMs in leveraging retrieved information or directly fine-tune LLMs to adapt to RAG scenarios. Although fine-tuning can yield better performance, it often compromises the LLMs' general generation capabilities by modifying their parameters. This limitation poses challenges in practical applications, especially when LLMs are already deployed, as parameter adjustments may affect their original functionality. To address this, we propose a novel method that involves learning scalable and pluggable virtual tokens for RAG. By maintaining the LLMs' original parameters and fine-tuning only the embeddings of these pluggable tokens, our approach not only enhances LLMs' performance but also preserves their general generation capabilities. Furthermore, we design several training strategies to improve the scalability, flexibility, and generalizability of our method. Comprehensive experiments across 12 question-answering tasks demonstrate the superiority of our approach.

Most existing graph clustering methods primarily focus on exploiting topological structure, often neglecting the "missing-half" node feature information, especially how these features can enhance clustering performance. This issue is further compounded by the challenges associated with high-dimensional features. Feature selection in graph clustering is particularly difficult because it requires simultaneously discovering clusters and identifying the relevant features for these clusters. To address this gap, we introduce a novel paradigm called "one node one model", which builds an exclusive model for each node and defines the node label as a combination of predictions for node groups. Specifically, the proposed "Feature Personalized Graph Clustering (FPGC)" method identifies cluster-relevant features for each node using a squeeze-and-excitation block, integrating these features into each model to form the final representations. Additionally, the concept of feature cross is developed as a data augmentation technique to learn low-order feature interactions. Extensive experimental results demonstrate that FPGC outperforms state-of-the-art clustering methods. Moreover, the plug-and-play nature of our method provides a versatile solution to enhance GNN-based models from the feature perspective.

Infusing artificial intelligence algorithms into production aerospace systems can be challenging due to costs, timelines, and a risk-averse industry. We introduce the Onboard Artificial Intelligence Research (OnAIR) platform, an open-source software pipeline and cognitive architecture tool that enables full life cycle AI research for on-board intelligent systems. We begin with a description and user walk-through of the OnAIR tool. Next, we describe four use cases of OnAIR for both research and deployed onboard applications, detailing their use of OnAIR and the benefits it provided to the development and function of each respective scenario. We conclude with remarks on future work, future planned deployments, and goals for the forward progression of OnAIR as a tool to enable a larger AI and aerospace research community.

Continual reinforcement learning (CRL) is the study of optimal strategies for maximizing rewards in sequential environments that change over time. This is particularly crucial in domains such as robotics, where the operational environment is inherently dynamic and subject to continual change. Nevertheless, research in this area has thus far concentrated on off-policy algorithms with replay buffers that are capable of amortizing the impact of distribution shifts. Such an approach is not feasible with on-policy reinforcement learning algorithms that learn solely from the data obtained from the current policy. In this paper, we examine the performance of proximal policy optimization (PPO), a prevalent on-policy reinforcement learning (RL) algorithm, in a classical CRL benchmark. Our findings suggest that the current methods are suboptimal in terms of average performance. Nevertheless, they demonstrate encouraging competitive outcomes with respect to forward transfer and forgetting metrics. This highlights the need for further research into continual on-policy reinforcement learning. The source code is available at https://github.com/Teddy298/continualworld-ppo.