AAAI 2025 Papers

Depth completion, predicting dense depth maps from sparse depth measurements, is an ill-posed problem requiring prior knowledge. Recent methods adopt learning-based approaches to implicitly capture priors, but the priors primarily fit in-domain data and do not generalize well to out-of-domain scenarios. To address this, we propose a zero-shot depth completion method composed of an affine-invariant depth diffusion model and test-time alignment. We use pre-trained depth diffusion models as depth prior knowledge, which implicitly understand how to fill in depth for scenes. Our approach aligns the affine-invariant depth prior with metric-scale sparse measurements, enforcing them as hard constraints via an optimization loop at test-time. Our zero-shot depth completion method demonstrates generalization across various domain datasets, achieving up to a 21% average performance improvement over the previous state-of-the-art methods while enhancing spatial understanding by sharpening scene details. We demonstrate that aligning a monocular affine-invariant depth prior with sparse metric measurements is a sufficient strategy to achieve domain-generalizable depth completion without relying on extensive training datasets.

LLMs obtain remarkable performance but suffer from hallucinations. Most research on detecting hallucination focuses on questions with short and concrete correct answers that are easy to check faithfulness. Hallucination detections for text generation with open-ended answers are more hard. Some researchers use external knowledge to detect hallucinations in generated texts, but external resources for specific scenarios are hard to access. Recent studies on detecting hallucinations in long texts without external resources conduct consistency comparison among multiple sampled outputs. To handle long texts, researchers split long texts into multiple facts and individually compare the consistency of each pair of facts. However, these methods (1) hardly achieve alignment among multiple facts; (2) overlook dependencies between multiple contextual facts. In this paper, we propose a graph-based context-aware (GCA) hallucination detection method for text generations, which aligns facts and considers the dependencies between contextual facts in consistency comparison. Particularly, to align multiple facts, we conduct a triple-oriented response segmentation to extract multiple knowledge triples. To model dependencies among contextual triples (facts), we construct contextual triples into a graph and enhance triples’ interactions via message passing and aggregating via RGCN. To avoid the omission of knowledge triples in long texts, we conduct an LLM-based reverse verification by reconstructing the knowledge triples. Experiments show that our model enhances hallucination detection and excels all baselines.

We present a novel, training-free approach to scene change detection. Our method leverages tracking models, which inherently perform change detection between consecutive frames of video by identifying common objects and detecting new or missing objects. Specifically, our method takes advantage of the change detection effect of the tracking model by inputting reference and query images instead of consecutive frames. Furthermore, we focus on the content gap and style gap between two input images in change detection, and address both issues by proposing adaptive content threshold and style bridging layers, respectively. Finally, we extend our approach to video, leveraging rich temporal information to enhance the performance of scene change detection. We compare our approach and baseline through various experiments. While existing train-based baseline tend to specialize only in the trained domain, our method shows consistent performance across various domains, proving the competitiveness of our approach.

Acquiring pairwise noisy-clean training data is challenging. Consequently, some self-supervised denoising methods utilize noisy image pairs as both input and target for network training. However, a major issue with these methods is the gap between the clean images of the input and target. In this paper, we achieve high-quality image denoising by reducing or even eliminating this gap. Our method requires no training data or prior knowledge of the noise distribution. It consists of two lightweight networks that can be trained using only a single noisy test image. Specifically, we propose a random mask-based downsampler that generates multiple pairs of downsampled noisy images, which are similar but distinct. These image pairs serve as the input for the first network, with the mean image of each pair used as the target. This initially reduces the gap between the clean images of the input and target. Particularly, in our method, the clean counterpart of the first network's target (i.e., the mean image) can be obtained. We then train a second network using the mean image as input and its clean counterpart as the target. This effectively eliminates the gap and achieves better denoising results. Extensive experiments demonstrate that our method outperforms in both denoising performance and efficiency.

Low-light image enhancement (LLIE) aims to improve visibility and signal-to-noise ratio in images captured under poor lighting conditions. While deep learning has shown promise in this domain, current approaches require extensive paired training data, limiting their practical utility. We present a novel framework that reformulates low-light image enhancement as a zero-shot inference problem using pre-trained latent diffusion models (LDMs), eliminating the need for task-specific training data. Our key insight is that the rich natural image priors encoded in LDMs can be leveraged to recover well-lit images through a carefully designed optimization process. To address the ill-posed nature of low-light degradation and the complexity of latent space optimization, our framework introduces an exposure-aware degradation module that adaptively models illumination variations and a principled latent regularization scheme with adaptive guidance that ensures both enhancement quality and natural image statistics. Experimental results demonstrate that our framework outperforms existing zero-shot methods across diverse real-world scenarios.

Large Visual Language Models (LVLMs) have achieved remarkable success in vision tasks. However, the significant differences between industrial and natural scenes make applying LVLMs challenging. Existing LVLMs rely on user-provided prompts to segment objects. This often leads to suboptimal performance due to the inclusion of irrelevant pixels. In addition, the scarcity of data also makes the application of LVLMs in industrial scenarios remain unexplored. To fill this gap, this paper proposes an open industrial dataset and a Refined Text-Visual Prompt (RTVP) for zero-shot industrial defect detection. First, this paper constructs the Multi-Modal Industrial Open Dataset (MMIO) containing 80K+ samples. MMIO contains diverse industrial categories, including 6 super categories and 18 subcategories. MMIO is the first large-scale multi-scenes pre-training dataset for industrial zero-shot learning, and provides valuable training data for open models in future industrial scenarios. Based on MMIO, this paper provides a RTVP specifically for industrial zero-shot tasks. RTVP has two significant advantages: First, this paper designs an expert-guided large model domain adaptation mechanism and designs an industrial zero-shot method based on Mobile-SAM, which enhances the generalization ability of large models in industrial scenarios. Second, RTVP automatically generates visual prompts directly from images and considers text-visual prompt interactions ignored by previous LVLM, improving visual and textual content understanding. RTVP achieves SOTA with 42.2% and 24.7% AP in zero-shot and closed scenes of MMIO.

Materials science text mining (MSTM), involving tasks like property extraction and synthesis action retrieval, is pivotal for advancing research by deriving critical insights from scientific literature. Descriptors, serving as essential task labels, often vary in meaning depending on researchers' usage purposes across different mining tasks. (e.g., 'Material' can refer to both synthesis components and participants in fuel cell experiment). This meaning difference makes it difficult for existing methods, fine-tuned to specific task, to handle the same descriptors in other tasks. To overcome above limitation, we propose MatDuck, a simple and effective approach for Zero-Shot MSTM by evoking material knowledge within Large Language Models (LLMs). Specifically, inspired by the Duck Typing principles in programming languages, we present a ClassDefinition-Style Descriptor generation method that evokes task-specific characteristics to address usage variation. Subsequently, we introduce code-style in-context learning for zero-shot tasks, reframing them into code to leverage LLMs' proficiency in code understanding. Extensive experiments on eight benchmark datasets demonstrate that MatDuck, as a plug-and-play approach, significantly improves the Zero-Shot MSTM performance of LLMs by an average of 11.3% across seven tasks.

As data and computational resources continue to expand, incorporating a variety of knowledge during the pre-training phase enhances large models, providing them with strong zero-shot capabilities. Due to the alignment of modal features by visual language models, zero-shot image captioning no longer necessitates pre-training on paired image-text labeled data, enabling accurate text description generation for images not encountered before. While recent research focuses on methods utilizing entity retrieval as anchors to bridge the gap between different modalities, these approaches often fall short of thoroughly analyzing the impact of entity retrieval recall on the zero-shot generation capabilities. To address this issue, we propose MERCap, a zero-shot image captioning method employing Multi-type Entity representation Retrieval. More specifically, we first approximate image representation using the CLIP representation of text and Gaussian noise to address the modality gap. Then, we train a GPT-2 decoder to reconstruct text using entities as hard prompts and CLIP representations as soft prompts. Additionally, we construct a domain-specific entity set, assigning multiple representations to each entity and refining their representation vectors through contrastive learning. During inference, we retrieve entities and input them into the decoder to generate corresponding captions. Extensive experiments validate that our approach is efficient, achieving a new state-of-the-art level in cross-domain captioning and demonstrating strong competitiveness in in-domain captioning compared to existing methods.

Score-based diffusion models have emerged as effective approaches for both conditional and unconditional generation. Still conditional generation is based on either a specific training of a conditional model or classifier guidance, which requires training a noise-dependent classifier, even when a classifier for uncorrupted data is given. We propose a method that, given a pre-trained unconditional score-based generative model, samples from the conditional distribution under arbitrary logical constraints, without requiring additional training. Differently from other zero-shot techniques, that rather aim at generating valid conditional samples, our method is designed for approximating the true conditional distribution. Firstly, we show how to manipulate the learned score in order to sample from an un-normalized distribution conditional on a user-defined constraint. Then, we define a flexible and numerically stable neuro-symbolic framework for encoding soft logical constraints. Combining these two ingredients we obtain a general, but approximate, conditional sampling algorithm. We further developed effective heuristics aimed at improving the approximation. Finally, we show the effectiveness of our approach in approximating conditional distributions for various types of constraints and data: tabular data, images and time series.

Generative AI presents transformative potential across various domains, from creative arts to scientific visualization. However, the utility of AI-generated imagery is often compromised by visual flaws, including anatomical inaccuracies, improper object placements, and misplaced textual elements. These imperfections pose significant challenges for practical applications. To overcome these limitations, we introduce Yuan, a novel framework that autonomously corrects visual imperfections in text-to-image synthesis. Yuan uniquely conditions on both the textual prompt and the segmented image, generating precise masks that identify areas in need of refinement without requiring manual intervention—a common constraint in previous methodologies. Following the automated masking process, an advanced inpainting module seamlessly integrates contextually coherent content into the identified regions, preserving the integrity and fidelity of the original image and associated text prompts. Through extensive experimentation on publicly available datasets such as ImageNet100 and Stanford Dogs, along with a custom-generated dataset, Yuan demonstrated superior performance in eliminating visual imperfections. Our approach consistently achieved higher scores in quantitative metrics, including NIQE, BRISQUE, and PI, alongside favorable qualitative evaluations. These results underscore Yuan's potential to significantly enhance the quality and applicability of AI-generated images across diverse fields.

Point clouds have become the preferred data format for a variety of tasks in 3D vision and graphics. However, raw point clouds often contain significant noise. This paper introduces the Adaptive Stop Denoising Network (ASDN), a novel approach aimed at restoring high-quality point clouds from noisy data. Our method is built upon a pivotal observation: during the denoising phase, high-noise points draw more focus from the network, which may suppress the points that have already been effectively denoised. This observation has led us to develop an adaptive strategy that ceases denoising already cleaned points to prevent over-denoising, while continuing to refine points that remain noisy. We employ a U-Net architecture complemented by an adaptive classifier, which utilizes a recoverability factor to assess the completion of denoising and make dynamic decisions about when to halt the process. Our method not only demonstrates superior noise removal efficiency but also preserves geometric details more effectively, reducing over- or under-denoising artifacts. Extensive experiments and evaluations demonstrate that our method outperforms the state-of-the-art both qualitatively and quantitatively.

Adverse clinical events related to unsafe care are among the top ten causes of death in the U.S. Accurate modeling and prediction of clinical events from electronic health records (EHRs) play a crucial role in patient safety enhancement. An example is modeling de facto care pathways that characterize common step-by-step plans for treatment or care. However, clinical event data pose several unique challenges, including the irregularity of time intervals between consecutive events, the existence of cycles, periodicity, multi-scale event interactions, and the high computational costs associated with long event sequences. Existing neural temporal point processes (TPPs) methods do not effectively capture the multi-scale nature of event interactions, which is common in many real-world clinical applications. To address these issues, we propose the cross-temporal-scale transformer (XTSFormer), specifically designed for irregularly timed event data. Our model consists of two vital components: a novel Feature-based Cycle-aware Time Positional Encoding (FCPE) that adeptly captures the cyclical nature of time, and a hierarchical multi-scale temporal attention mechanism, where different temporal scales are determined by a bottom-up clustering approach. Extensive experiments on several real-world EHR datasets show that our XTSFormer outperforms multiple baseline methods.

Cotton is a critical agricultural product and industrial raw material, playing a key role in the national economies and people's living conditions, particularly in developing countries. However, cotton picking and processing often result in the contamination with various foreign fibers, such as hair, hemp rope, plastic film, and polypropylene rope. These contaminants are difficult to remove during textile processing and tend to break into small fragments, significantly reducing the quality of cotton products and negatively impacting the cotton industry. In this paper, we present an AI-enabled hardware-software integrated system--XCotton, for identifying and removing foreign fibers. Our system has been deployed in actual cotton production environments in the multiple regions in China, Central Asia, and Africa. XCotton achieves a cleaning efficiency of 1000kg/h, representing a 43% improvement, with only 14 kWh energy consumption (63% less). Moreover, XCotton brings significant business values to its manufacturer and clients. XCotton not only enhances the quality of cotton products but also contributes to the value-adding and upgrading of the cotton industry in developing regions, supporting economic growth and improving living conditions.

Chain-of-thought (CoT) has emerged as a powerful technique to elicit reasoning in large language models and improve a variety of downstream tasks. CoT mainly demonstrates excellent performance in English, but its usage in low-resource languages is constrained due to poor language generalization. To bridge the gap among different languages, we propose a cross-lingual instruction fine-tuning framework (xCoT) to transfer knowledge from high-resource languages to low-resource languages. Specifically, the multilingual instruction training data (xCoT-Instruct) is created to encourage the semantic alignment of multiple languages. We introduce cross-lingual in-context few-shot learning (xICL) to accelerate multilingual agreement in instruction tuning, where some fragments of source languages in examples are randomly substituted by their counterpart translations of target languages. During multilingual instruction tuning, we adopt the randomly online CoT strategy to enhance the multilingual reasoning ability of the large language model by first translating the query to another language and then answering in English. To further facilitate the language transfer, we leverage the high-resource CoT to supervise the training of low-resource languages with cross-lingual distillation. Experimental results demonstrate the superior performance of xCoT in reducing the gap among different languages, highlighting its potential to reduce the cross-lingual gap.

AI systems make decisions in physical environments through primitive actions or affordances that are accessed via API calls. While deploying AI agents in the real world involves numerous high-level actions, existing embodied simulators offer a limited set of domain-salient APIs. This naturally brings up the questions: how many primitive actions (APIs) are needed for a versatile embodied agent, and how should they look like? We explore this via a thought experiment: assuming that wikiHow tutorials cover a wide variety of human-written tasks, what is the space of APIs needed to cover these instructions? We propose a framework to iteratively induce new APIs by grounding wikiHow instruction to situated agent policies. Inspired by recent successes in large language models (LLMs) for embodied planning, we propose a few-shot prompting to steer GPT-4 to generate Pythonic programs as agent policies and bootstrap a universe of APIs by 1) reusing a seed set of APIs; and then 2) fabricate new API calls when necessary. The focus of this thought experiment is on defining these APIs rather than their excitability. We apply the proposed pipeline on instructions from wiki- How tutorials. On a small fraction (0.5%) of tutorials, we induce an action space of 300+ APIs necessary for capturing the rich variety of tasks in the physical world. A detailed automatic and human analysis of the induction output reveals that the proposed pipeline enables effective reuse and creation of APIs. Moreover, a manual review revealed that existing simulators support only a small subset of the induced APIs (9 of the top 50 frequent APIs), motivating the development of action-rich embodied environments.

The Multi-modal Large Language Models (MLLMs) with extensive world knowledge have revitalized autonomous driving, particularly in reasoning tasks within perceivable regions. However, when faced with perception-limited areas (dynamic or static occlusion regions), MLLMs struggle to effectively integrate perception ability with world knowledge for reasoning. These perception-limited regions can conceal crucial safety information, especially for vulnerable road users. In this paper, we propose a framework, which aims to improve autonomous driving performance under perception-limited conditions by enhancing the integration of perception capabilities and world knowledge. Specifically, we propose a plug-and-play instruction-guided interaction module that bridges modality gaps and significantly reduces the input sequence length, allowing it to adapt effectively to multi-view video inputs. Furthermore, to better integrate world knowledge with driving-related tasks, we have collected and refined a large-scale multi-modal dataset that includes 2 million natural language QA pairs, 1.7 million grounding task data. To evaluate the model’s utilization of world knowledge, we introduce an object-level risk assessment dataset comprising 200K QA pairs, where the questions necessitate multi-step reasoning leveraging world knowledge for resolution. Extensive experiments validate the effectiveness of our proposed method.

As Artificial Intelligence (AI) continues to integrate into more aspects of society, equipping younger generations with foundational AI knowledge becomes increasingly critical. This paper presents Word2Vec4Kids (W2V4K), an interactive application designed to familiarize middle school students with word embeddings, a key aspect of Natural Language Processing (NLP). W2V4K leverages the Word2Vec model, allowing students to explore word associations, similarity, and vector arithmetic through engaging game modes. The application was tested with 38 middle school students aged 11-14 at a Science Technology Engineering Math (STEM)-focused charter school. Data were collected on students' interactions with the application, including screen recordings, audio, and survey responses. Results demonstrated that W2V4K effectively introduces NLP concepts to students. Qualitative observations revealed high levels of engagement with students expressing excitement and curiosity about word relationships. As they progressed through the game modes, students showed increasing confidence in predicting word associations, brainstorming relevant words, and connecting the concepts to real-world applications. Quantitative data from post-interaction surveys indicated positive learning outcomes with 44.5% of students achieving perfect scores on concept-related items. Additionally, students demonstrated an ability to critically think about language representation. This study suggests that W2V4K provides an effective and engaging method for introducing NLP concepts to middle school students, contributing to the broader goal of enhancing AI literacy among younger generations.

Decision trees are a classic model for summarizing and classifying data. To enhance interpretability and generalization properties, it has been proposed to favor small decision trees. Accordingly, in the minimum-size decision tree training problem (MSDT), the input is a set of training examples in $\mathbb{R}^d$ with class labels and we aim to find a decision tree that classifies all training examples correctly and has a minimum number of nodes. MSDT is NP-hard and therefore presumably not solvable in polynomial time. Nevertheless, a promising algorithmic paradigm called witness trees which solves MSDT efficiently if the solution tree is small has been developed. In this work, we test this paradigm empirically. We provide an implementation, augment it with extensive heuristic improvements, and scrutinize it on standard benchmark instances. The augmentations achieve a mean 324-fold (median 84-fold) speedup over the naive implementation. Compared to the state of the art they achieve a mean 32-fold (median 7-fold) speedup over the dynamic programming based MurTree solver and a mean 61-fold (median 25-fold) speedup over SAT-based implementations. As a theoretical result we obtain an improved worst-case running-time bound for MSDT.

Traditional multi-agent path finding (MAPF) methods try to compute entire collision free start-goal paths, with several algorithms offering completeness guarantees. However, computing partial paths offers significant advantages including faster planning, adaptability to changes, and enabling decentralized planning. Methods that compute partial paths employ a "windowed" approach and only try to find collision free paths for a limited timestep horizon. While this improves flexibility, this adaptation introduces incompleteness; all existing windowed approaches can become stuck in deadlock or livelock. Our main contribution is to introduce our framework, WinC-MAPF, for Windowed MAPF that enables completeness. Our framework leverages heuristic update insights from single-agent real-time heuristic search algorithms and agent independence ideas from MAPF algorithms. We also develop Single-Step Conflict Based Search (SS-CBS), an instantiation of this framework using a novel modification to CBS. We show how SS-CBS, which only plans a single step and updates heuristics, can effectively solve tough scenarios where existing windowed approaches fail.