AAAI 2025 Papers

Cross-Domain Few-Shot Learning (CDFSL) requires the model to transfer knowledge from the data-abundant source domain to data-scarce target domains for fast adaptation, where the large domain gap makes CDFSL a challenging problem. Masked Autoencoder (MAE) excels in effectively using unlabeled data and learning image’s global structures, enhancing model generalization and robustness. However, in the CDFSL task with significant domain shifts, we find MAE even shows lower performance than the baseline supervised models. In this paper, we first delve into this phenomenon for an interpretation. We find that MAE tends to focus on low-level domain information during reconstructing pixels while changing the reconstruction target to token features could mitigate this problem. However, not all features are beneficial, as we then find reconstructing high-level features can hardly improve the model’s transferability, indicating a trade-off between filtering domain information and preserving the image’s global structure. In all, the reconstruction target matters for the CDFSL task. Based on the above findings and interpretations, we further propose Domain-Agnostic Masked Image Modeling (DAMIM) for the CDFSL task. DAMIM includes an Aggregated Feature Reconstruction module to automatically aggregate features for reconstruction, with balanced learning of domain-agnostic information and images’ global structure, and a Lightweight Decoder module to further benefit the encoder’s generalizability. Experiments on four CDFSL datasets demonstrate that our method achieves state-of-the-art performance.

The purpose of partial multi-label feature selection is to select the most representative feature subset, where the data comes from partial multi-label datasets that have label ambiguity issues. For label disambiguation, previous methods mainly focus on utilizing the information inside the labels and the relationship between the labels and features. However, the information existing in the feature space is rarely considered, especially in partial multi-label scenarios where the noises is considered to be concentrated in the label space while the feature information is correct. This paper proposes a method based on latent space alignment, which uses the information mined in feature space to disambiguate in latent space through the structural consistency between labels and features. In addition, previous methods overestimate the consistency of features and labels in the latent space after convergence. We comprehensively consider the similarity of latent space projections to feature space and label space, and propose new feature selection term. This method also significantly improves the positive label identification ability of the selected features. Comprehensive experiments demonstrate the superiority of the proposed method.

Multi-label class-incremental learning (MLCIL) is essential for real-world multi-label applications, allowing models to learn new labels while retaining previously learned knowledge continuously. However, recent MLCIL approaches can only achieve suboptimal performance due to the oversight of the positive-negative imbalance problem, which manifests at both the label and loss levels because of the task-level partial label issue. The imbalance at the label level arises from the substantial absence of negative labels, while the imbalance at the loss level stems from the asymmetric contributions of the positive and negative loss parts to the optimization. To address the issue above, we propose a Rebalance framework for both the Loss and Label levels (RebLL), which integrates two key modules: asymmetric knowledge distillation (AKD) and online relabeling (OR). AKD is proposed to rebalance at the loss level by emphasizing the negative label learning in classification loss and down-weighting the contribution of overconfident predictions in distillation loss. OR is designed for label rebalance, which restores the original class distribution in memory by online relabeling the missing classes. Our comprehensive experiments on the PASCAL VOC and MS-COCO datasets demonstrate that this rebalancing strategy significantly improves performance, achieving new state-of-the-art results even with a vanilla CNN backbone.

This paper considers the challenges Large Language Models (LLMs) face when reasoning over text that includes information involving uncertainty explicitly quantified via probability values. This type of reasoning is relevant to a variety of contexts ranging from everyday conversations to medical decision-making. Despite improvements in the mathematical reasoning capabilities of LLMs, they still exhibit significant difficulties when it comes to probabilistic reasoning. To deal with this problem, we introduce the Bayesian Linguistic Inference Dataset (BLInD), a new dataset specifically designed to test the probabilistic reasoning capabilities of LLMs. We use BLInD to find out the limitations of LLMs for tasks involving probabilistic reasoning. In addition, we present several prompting strategies that map the problem to different formal representations, including Python code, probabilistic algorithms, and probabilistic logical programming. We conclude by providing an evaluation of our methods on BLInD and an adaptation of a causal reasoning question-answering dataset. Our empirical results highlight the effectiveness of our proposed strategies for multiple LLMs.

Reasoning about the causes behind observations is crucial to the formalization of rationality. While extensive research has been conducted on root cause analysis, most studies have predominantly focused on deterministic settings. In this paper, we investigate causation in more realistic nondeterministic domains, where the agent does not have any control on and may not know the choices that are made by the environment. We build on recent preliminary work on actual causation in the nondeterministic situation calculus to formalize more sophisticated forms of reasoning about actual causes in such domains. We investigate the notions of “Certainly Causes” and “Possibly Causes” that enable the representation of actual cause for agent actions in these domains. We then show how regression in the situation calculus can be extended to reason about such notions of actual causes.

AI has immense potential for positive social impact, including in domains ranging from conservation to health. However, it can be challenging to account for human collaborations and real-world uncertainties when deploying such systems, which can lead to critical errors. Therefore, my research focuses on developing new methods in multi-agent systems and machine learning, including methods for participatory design of AI, human-AI collaboration, and uncertainty quantification, to develop safe, impactful AI systems, particularly in the domains of water conservation and reproductive health.

Deep denoising models require extensive real-world training data, which is challenging to acquire. Current noise synthesis techniques struggle to accurately model complex noise distributions. We propose a novel Realistic Noise Synthesis Diffusor (RNSD) method using diffusion models to address these challenges. By encoding camera settings into a time-aware camera-conditioned affine modulation (TCCAM), RNSD generates more realistic noise distributions under various camera conditions. Additionally, RNSD integrates a multi-scale content-aware module (MCAM), enabling the generation of structured noise with spatial correlations across multiple frequencies. We also introduce Deep Image Prior Sampling (DIPS), a learnable sampling sequence based on depth image prior, which significantly accelerates the sampling process while maintaining the high quality of synthesized noise. Extensive experiments demonstrate that our RNSD method significantly outperforms existing techniques in synthesizing realistic noise under multiple metrics and improving image denoising performance.

Recently, the success of text-to-image synthesis has greatly advanced the development of identity customization techniques, whose main goal is to produce realistic identity-specific photographs based on text prompts and reference face images. However, it is difficult for existing identity customization methods to simultaneously meet the various requirements of different real-world applications, including the identity fidelity of small face, the control of face location, pose and expression, as well as the customization of multiple persons. To this end, we propose a scale-robust and fine-controllable method, namely RealisID, which learns different control capabilities through the cooperation between a pair of local and global branches. Specifically, by using cropping and up-sampling operations to filter out face-irrelevant information, the local branch concentrates the fine control of facial details and the scale-robust identity fidelity within the face region. Meanwhile, the global branch manages the overall harmony of the entire image. It also controls the face location by taking the location guidance as input. As a result, RealisID can benefit from the complementarity of these two branches. Finally, by implementing our branches with two different variants of ControlNet, our method can be easily extended to handle multi-person customization, even only trained on single-person datasets. Extensive experiments and ablation studies indicate the effectiveness of RealisID and verify its ability in fulfilling all the requirements mentioned above.

In recent years, diffusion models have revolutionized visual generation, outperforming traditional frameworks like Generative Adversarial Networks (GANs). However, generating images of humans with realistic semantic parts, such as hands and faces, remains a significant challenge due to their intricate structural complexity. To address this issue, we propose a novel post-processing solution named RealisHuman. The RealisHuman framework operates in two stages. First, it generates realistic human parts, such as hands or faces, using the original malformed parts as references, ensuring consistent details with the original image. Second, it seamlessly integrates the rectified human parts back into their corresponding positions by repainting the surrounding areas to ensure smooth and realistic blending. The RealisHuman framework significantly enhances the realism of human generation, as demonstrated by notable improvements in both qualitative and quantitative metrics.

We introduce RealPortrait, a framework based on Diffusion Transformers (DiT), designed to generate highly expressive and visually appealing portrait animations. Given a static portrait image, our method can transfer complex facial expressions and head pose movements extracted from a driving video onto the portrait, transforming it into a lifelike video. Specifically, we exploit the robust spatial-temporal modeling capabilities of DiT, enabling the generation of portrait videos that maintain high-fidelity visual details and ensure temporal coherence. In contrast to conventional image-to-video generation frameworks that necessitate a separate reference network, we incorporate an efficient reference attention within the DiT backbone, thereby obviating the computational overhead and achieving superior reference appearance preservation. Concurrently, we integrate a parallel ControlNet to precisely regulate intricate facial expressions and head poses. Diverging from prior methods that utilize explicit sparse motion representations, such as facial landmarks or 3DMM coefficients, we adopt a dense implicit motion representation as the control guidance. This implicit motion representation excels in capturing nuanced emotional facial expressions and subtle non-rigid dynamics of the lips. To further enhance the generalization capability of the model, we augment the training dataset by incorporating a substantial volume of facial image data through random crop augmentation. This strategy ensures the model's robustness across a wide variety of facial appearances and expressions. Empirical evaluations demonstrate that RealPortrait excels in generating portrait animations with highly-realistic quality and exceptional temporal coherence in appearance retention.

We introduce a biologically plausible RL framework for solving tasks in partially observable Markov decision processes (POMDPs). The proposed algorithm combines three integral parts: (1) A Meta-RL architecture, resembling the mammalian basal ganglia; (2) A biologically plausible reinforcement learning algorithm, exploiting temporal difference learning and eligibility traces to train the policy and the value-function; (3) An online automatic differentiation algorithm for computing the gradients with respect to parameters of a shared recurrent network backbone. Our experimental results show that the method is capable of solving a diverse set of partially observable reinforcement learning tasks. The algorithm we call real-time recurrent reinforcement learning (RTRRL) serves as a model of learning in biological neural networks, mimicking reward pathways in the basal ganglia.

The increasing demand for real-time analysis in video streaming has driven significant advancements in object detection and motion prediction. This paper presents SkelAI, an innovative application that combines YOLOv8, OpenCV, OpenAI API, and our own innovative algorithms to achieve real-time object detection and medial axis skeletonization tailored explicitly for live video streaming environments. In addition, SkelAI integrates AI-generated image capabilities through the DALL-E 3 model, enabling the extraction of skeletons from synthetic content that simulates streaming scenarios. The application supports exporting skeleton data in PyTorch-compatible formats, facilitating the training of sequence predicting deep learning models. Comprehensive evaluations demonstrate SkelAI’s enhanced accuracy, efficiency, and versatility compared to existing tools, underscoring its potential applications in digital animation, biomechanical research and robotics, human-computer interaction, and video compression within streaming platforms.

Real-time Monte Carlo (MC) ray tracing with low sampling rates demands a denoising algorithm that adeptly balances the trade-off between quality and efficiency. Previous works have paid much attention on designing delicate denoising architecture while ignoring model compression. In this work, we present a render-aware knowledge distillation (RAKD) framework, specifically designed for Monte Carlo denoising. We meticulously delineate the Knowledge Distillation (KD) process within RAKD, emphasizing three pivotal techniques: the strategic incorporation of an auxiliary unlabeled dataset, the integration of adversarial learning through generative adversarial network (GAN), and the application of parameter transfer for robust model initialization. These approaches are harmoniously combined to distill knowledge effectively, enabling our student model to adeptly strike a balance between preserving high-frequency details and reducing low-frequency noise. Finally, our results demonstrate that RAKD achieves state-of-the-art quality while upholding real-time performance, successfully tackling the computational constraints faced by resource-limited devices.

Precise measurements from sensors are crucial, but data is usually collected from low-cost, low-tech systems, which are often inaccurate. Thus, they require further calibrations. To that end, we first identify three requirements for effective calibration under practical low-tech sensor conditions. Based on the requirements, we develop a model called TESLA, Transformer for effective sensor calibration utilizing logarithmic-binned attention. TESLA uses a high-performance deep learning model, Transformers, to calibrate and capture non-linear components. At its core, it employs logarithmic binning, to minimize attention complexity. TESLA achieves consistent real-time calibration, even with longer sequences and finer-grained time series in hardware-constrained systems. Experiments show that TESLA outperforms existing novel deep learning and newly crafted linear models in accuracy, calibration speed, and energy efficiency.

Recommender systems used in online platforms can drive users to consume content continuously in an attempt to maximize satisfaction. Such engagement is invariably broken due to more pressing work, alternate pursuits, distractions or fatigue. Recommender systems need to ensure the continuity of experience when the user joins back. Session-based recommender systems typically create different sessions based on a fixed time interval (θ), often resulting in creation of a separate session when the user gets off the platform temporarily. When the user joins back, session-based recommender systems are likely to recommend content different than what they would have in case the earlier session had continued. This may cause dissatisfaction given that there is a difference in the predicted world model of the user, i.e. the expectation from the last session, and the observed one, i.e. the recommendations. To handle this problem, we propose the creation of content-driven sessions instead of time-driven sessions. In our setting, a session continues while a single item category dominates in the user-item interactions. A new session is created when a different item category begins to dominate. The proposed content-driven method also solves the long-standing problem of deciding the optimal value of time threshold (θ) for defining the time-based session. We report that the proposed method outperforms existing SOTA methodologies set by time-based sessions by a large margin in terms of recommendation performance on multiple datasets.

Despite the impressive progress of multimodal generative models, video-to-audio generation still suffers from limited performance and limits the flexibility to prioritize sound synthesis for specific objects within the scene. Conversely, text-to-audio generation methods generate high-quality audio but pose challenges in ensuring comprehensive scene depiction and time-varying control. To tackle these challenges, we propose a novel video-and-text-to-audio generation method, called ReWaS, where video serves as a conditional control for a text-to-audio generation model. Especially, our method estimates the structural information of sound (namely, energy) from the video while receiving key content cues from a user prompt. We employ a well-performing text-to-audio model to consolidate the video control, which is much more efficient for training multimodal diffusion models with massive triplet-paired (audio-video-text) data. In addition, by separating the generative components of audio, it becomes a more flexible system that allows users to freely adjust the energy, surrounding environment, and primary sound source according to their preferences. Experimental results demonstrate that our method shows superiority in terms of quality, controllability, and training efficiency.

The interdisciplinary field of chemistry and artificial intelligence (AI) is an active area of research aimed at accelerating scientific discovery. Large language Models (LLMs) have shown significant promise in biochemical tasks, especially the molecule caption translation, which aims to align between molecules and natural language texts. However, existing works mainly focus on single molecules, while alignment between chemical reactions and natural language text remains largely unexplored. Additionally, the description of reactions is an essential part in biochemical patents and literature, and research on this aspect not only can help better understand chemical reactions but also promote research on automating chemical synthesis and retrosynthesis. In this work, we propose \textbf{ReactGPT}, a framework aiming to bridge the gap between chemical reaction and text. ReactGPT allows a new task: reaction captioning, by adapting LLMs to learn reaction-text alignment from context examples via In-Context Tuning. Specifically, ReactGPT jointly leverages a Fingerprints-based Reaction Retrieval module, a Domain-Specific Prompt Design module, and a two-stage In-Context Tuning module. We evaluate the effectiveness of ReactGPT on reaction captioning and experimental procedure prediction, both of these tasks can reflect the understanding of chemical reactions. Experimental results show that compared to previous models, ReactGPT exhibits competitive capabilities in resolving chemical reactions and generating high-quality text with correct structure.

Existing local model-agnostic explanation techniques are ineffective for machine learning models that consider inputs of variable lengths, as they do not consider temporal information embedded in these models. To address this limitation, we propose ReX, a general framework for incorporating temporal information in these techniques. Our key insight is that these techniques typically learn a model surrogate by sampling model inputs and outputs, and we can incorporate temporal information in a uniform way by only changing the sampling process and the surrogate features. We instantiate our approach on three popular explanation techniques: Anchors, LIME, and Kernel SHAP. To evaluate the effectiveness of ReX, we apply our approach to six models in three different tasks. Our evaluation results demonstrate that our approach 1) significantly improves the fidelity of explanations, making model-agnostic techniques outperform a state-of-the-art model-specific technique on its target model, and 2) helps end users better understand the models' behaviors.

Generation of 3D human motion holds significant importance in the creative industry. While recent notable advances have been made in generating common motions, existing methods struggle to generate diverse and rare motions due to the complexity of motions and limited training data. This work introduces ReMoGPT, a unified motion-language generative model that solves a wide range of motion-related tasks by incorporating a multi-modal retrieval mechanism into the generation process to address the limitations of existing models, namely diversity and generalizability. We propose to focus on body-part-level motion features to enable fine-grained text-motion retrieval and locate suitable references from the database to conduct generation. Then, the motion-language generative model is trained with prompt-based question-and-answer tasks designed for different motion-relevant problems. We incorporate the retrieved samples into the prompt, and then perform instruction tuning of the motion-language model, to learn from task feedback and produce promising results with the help of fine-grained multi-modal retrieval. Extensive experiments validate the efficacy of ReMoGPT, showcasing its superiority over existing state-of-the-art methods. The framework performs well on multiple motion tasks, including motion retrieval, generation, and captioning.