AAAI 2025 Papers

Location-Based Social Networks (LBSNs) offer a rich dataset of user activity at Points-of-Interest (POIs), making next POI recommendation a key task. Traditional algorithms face challenges due to broad searching scopes, affecting recommendation accuracy. Users tend to visit nearby POIs and show temporal concentration in their activities, reflecting personalized spatio-temporal clustering. However, individual user data may be insufficient to capture these clustering effects for personalized recommendations. In this paper, we propose an integrated Personalized Spatio-Temporal Clustering Model (iPCM) for next POI recommendation. The model learns this kind of personalized spatio-temporal clustering effect by using global historical trajectory data in conjunction with user feature embeddings. It integrates the features of personalized spatio-temporal clustering with the user's trajectory, and completes the user's POI recommendation through a Transformer encoding and MLP decoding. To enhance the accuracy of predictions, we add a module of probability adjustment. The experimental results on multiple datasets show that with the help of personalized spatio-temporal clustering, the proposed iPCM is superior to existing methods in various evaluation metrics.

Long-sequence precipitation forecasting is critical for both meteorological science and smart city applications. The primary objective of this task is to predict future radar echo sequences, which provide high resolution and timely references for atmospheric precipitation distribution based on current observations. However, the chaotic nature of precipitation systems poses significant challenges in extending reliable forecast horizons. Most existing methods struggle with accuracy and clarity when extended to long-sequence predictions, such as three-hour forecasts. This is primarily due to the insufficiency of spatio-temporal information within a single modality over time. In this paper, we propose a cascading forecasting framework that adaptively extracts and integrates multimodal spatio-temporal information to support accurate and realistic long-sequence radar forecasting. Our framework includes a temporal adaptive predictor and a flow-based precipitation distribution adaptor. The predictor utilizes a multi-branch encoder-decoder architecture. This design allows it to extract meteorological sequences from multiple sources at varying scales, resulting in an initial global precipitation estimate. The core component is a carefully designed cross-attention module with a temporal adaptive layer to enhance multi-modality alignment. The initial estimate is then refined by the flow-based adaptor, which adjusts the prediction to match the target precipitation distribution, enhancing local details and correcting extreme precipitation patterns. We validated our method using real multi-source dataset for long-sequence forecasting, and the experimental results demonstrate that our approach outperforms existing state-of-the-art methods.

Unsupervised semantic segmentation algorithms aim to identify meaningful semantic groups without annotations. Recent approaches leveraging self-supervised transformers as pre-training backbones have successfully obtained high-level dense features that effectively express semantic coherence. However, these methods often overlook local semantic coherence and low-level features such as color and texture. We propose integrating low-level visual cues to complement high-level visual cues derived from self-supervised pre-training branches. Our findings indicate that low-level visual cues provide a more coherent recognition of color-texture aspects, ensuring the continuity of spatial structures within classes. This insight led us to develop IL2Vseg, an unsupervised semantic segmentation method that leverages the complementation of low-level visual cues. The core of IL2Vseg is a spatially-constrained fuzzy clustering algorithm based on color affinities, which preserves the intra-class affinity of spatially-adjacent and similarly-colored pixels in low-level visual cues. Additionally, to effectively couple low-level and high-level visual cues, we introduce a feature similarity loss function to optimize the feature representation of fused visual cues. To further enhance consistent feature learning, we incorporate contrast loss functions based on color invariance and luminosity invariance, which improve the learning of features from different semantic categories. Extensive experiments on multiple datasets, including COCO-Stuff-27, Cityscapes, Potsdam, and MaSTr1325, demonstrate that IL2Vseg achieves state-of-the-art results.

The significance of Temporal Knowledge Graphs (TKGs) in Artificial Intelligence (AI) lies in their capacity to incorporate time-dimensional information, support complex reasoning and prediction, optimize decision-making processes, enhance the accuracy of recommendation systems, promote multimodal data integration, and strengthen knowledge management and updates. This provides a robust foundation for various AI applications. To effectively learn and apply both static and dynamic temporal patterns for reasoning, a range of embedding methods and large language models (LLMs) have been proposed in the literature. However, these methods often rely on a single underlying embedding space, whose geometric properties severely limit their ability to model intricate temporal patterns, such as hierarchical and ring structures. To address this limitation, this paper proposes embedding TKGs into projective geometric space and leverages LLMs technology to extract crucial temporal node information, thereby constructing the 5EL model. By embedding TKGs into projective geometric space and utilizing Möbius Group transformations, we effectively model various temporal patterns. Subsequently, LLMs technology is employed to process the trained TKGs. We adopt a parameter-efficient fine-tuning strategy to align LLMs with specific task requirements, thereby enhancing the model's ability to recognize structural information of key nodes in historical chains and enriching the representation of central entities. Experimental results on five advanced TKG datasets demonstrate that our proposed 5EL model significantly outperforms existing models.

The strategic behavior of users is significantly influenced by their hidden information such as private valuations, risk preferences, and price sensitivities. Contextual behavioral model learning refers to learning the dependence of users' hidden information on their observable context information. While many existing studies use offline data to learn contextual behavioral models, we study how to design sequential experiments to collect the most informative user behavioral data for learning. We propose a basic inference-then-design method. In each experimental period, it infers a probabilistic contextual behavioral model using historical experimental data, and then designs the new experiment to maximize the gain of information about the probabilistic model. We further improve the basic method in two aspects. First, we improve the inference step by specifying a more informative prior for learning the probabilistic contextual behavioral model. Second, we integrate the inference and design steps instead of conducting them separately. Our rigorous theoretic analysis reveals that the optimization objective of the inference step can be modified to account for the downstream experimental design step. Numerical experiments show that our methods lead to more effective experiments, i.e., the collected experimental data can help in learning a more accurate behavioral model.

Graph Neural Networks (GNNs) have recently achieved significant success in several graph-related tasks. However, traditional GNNs and their variants are constantly limited by the implicit homophily, assuming neighboring nodes belong to the same class. This results in weak performance on heterophilic graphs where most nodes are linked to neighbors of different classes. Despite the numerous attempts to adequately deal with heterophily, most methods still use the uniform propagation aggregation mechanism. In this paper, we argue that identifying neighbors with different class labels and exploiting them individually is crucial for heterophilic GNNs. We then propose a simple and efficient novel co-training approach, EG-GCN, which uses group aggregation to handle homophilic and heterophilic neighbors separately. In EG-GCN, we first use an edge discriminator to classify edges and split the neighborhood of every node into two parts. We then apply group graph convolution to the divided neighborhoods to obtain node representations. During training, we continuously optimize the edge discriminator to improve neighborhood partition and use the node classification results to identify highly confident unlabeled nodes to expand the edge training set. This co-training strategy enables both components to enhance each other mutually. Extensive experiments demonstrate that EG-GCN significantly outperforms the state-of-the-art approaches.

Explanation for deep learning models on time series classification (TSC) tasks is an important and challenging problem. Most existing approaches use attribution maps to explain outcomes. However, they have limitations in generating explanations that are well-aligned with humans's perceptions. Recently LIME-based approaches provide a more meaningful explanation via segmenting the data. However, these approaches are still suffering from the processes of segment generations and evaluations. In this paper, we propose a novel time series explanation approach called InteDisUX to overcome these problems. Our technique utilizes the segment-level integrated gradient (SIG) for calculating importance scores for an initial set of small and equal segments before iteratively merge two consecutive ones to create better explanations under a unique greedy strategy guided by two new proposed metrics including discrimination and faithfulness gains. By this way, our method does not depend on predefined segments like others while being robusts to instability, poor local fidelity and data imbalance like LIME-based methods. Furthermore, InteDisUX is the first work to use the model's information to improve the set of segments} for time series explanation. Extensive experiments show that our method outperforms LIME-based ones in 12 datasets in terms of faithfulness and 8/12 datasets in terms of robustness.

Motion planning is a critical module in autonomous driving, with the primary challenge of uncertainty caused by interactions with other participants. As most previous methods treat prediction and planning as separate tasks, it is difficult to model these interactions. Furthermore, since the route path navigates ego vehicles to a predefined destination, it provides relatively stable intentions for ego vehicles and helps constrain uncertainty. On this basis, we construct Int2Planner, an Intention-based Integrated motion Planner achieves multi-modal planning and prediction. Instead of static intention points, Int2Planner utilizes route intention points for ego vehicles and generates corresponding planning trajectories for each intention point to facilitate multi-modal planning. The experiments on the private dataset and the public nuPlan benchmark show the effectiveness of route intention points, and Int2Planner achieves state-of-the-art performance. We also deploy it in real-world vehicles and have conducted autonomous driving for hundreds of kilometers in urban areas. It further verifies that Int2Planner can continuously interact with the traffic environment.

Open-set speaker recognition is to identify whether the voices are from the same speaker. One challenge of speaker recognition is collecting large amounts of high-quality data. Based on the promising results of image classification, one intuitively feasible solution is semi-supervised learning (SSL) which uses confidence thresholds to assign pseudo labels for unlabeled data. However, we empirically demonstrated that applying SSL methods to speaker recognition is non-trivial. These methods focus solely on inter-class discrepancy as thresholds to select pseudo labels, overlooking intra-class compactness, which is particularly important for open-set speaker recognition tasks. Motivated by this, we propose Int*-Match, a semi-supervised speaker recognition method selecting reliable pseudo labels with intra-class compactness and inter-class discrepancy for speaker recognition. In particular, we use the inter-class discrepancy of labeled data as the threshold for pseudo-label selection and adjust the threshold based on the intra-class compactness of the pseudo labels dynamically and adaptively. Our systematic experiments demonstrate the superiority of Int*-Match, presenting an outstanding Equal Error Rate (EER) of 1.00% on the VoxCeleb1 original test set, which is merely 0.06% below the performance achieved by fully supervised learning.

Large Multimodal Models (LMMs) have significantly progressed by extending large language models. Building on this progress, the latest developments in LMMs demonstrate the ability to generate dense pixel-wise segmentation by integrating segmentation models. Despite the innovations, existing works’ textual responses and segmentation masks remain at the instance level, showing limited ability to perform fine-grained understanding and segmentation even provided with detailed textual cues. To overcome this limitation, we introduce a Multi-Granularity Large Multimodal Model (MGLMM), which is capable of seamlessly adjusting the granularity of Segmentation and Captioning (SegCap) following user instructions, from panoptic SegCap to fine-grained SegCap. We name such a new task Multi-Granularity Segmentation and Captioning (MGSC). Observing the lack of a benchmark for model training and evaluation over the MGSC task, we establish a benchmark with aligned masks and captions in multi-granularity using our customized automated annotation pipeline. This benchmark comprises 10K images and more than 30K image-question pairs. We will release our dataset along with the implementation of our automated dataset annotation pipeline for further research. Besides, we propose a novel unified SegCap data format to unify heterogeneous segmentation datasets; it effectively facilitates learning to associate object concepts with visual features during multi-task training. Extensive experiments demonstrate that our MGLMM excels at tackling more than eight downstream tasks and achieves state-of-the-art performance in MGSC, GCG, image captioning, referring segmentation, multiple/empty segmentation, and reasoning segmentation. The great properties and versatility of MGLMM underscore its potential impact on advancing multimodal research.

Enabling humanoid robots to perform long-horizon mobile manipulation planning in real-world environments based on embodied perception and comprehension abilities has been a longstanding challenge. With the recent rise of large language models (LLMs), there has been a notable increase in the development of LLM-based planners. These approaches either utilize human-provided textual representations of the real world or heavily depend on prompt engineering to extract such representations, lacking the capability to quantitatively understand the environment, such as determining the feasibility of manipulating objects. To address these limitations, we present the Instruction-Augmented Long-Horizon Planning (IALP) system, a novel framework that employs LLMs to generate feasible and optimal actions based on real-time sensor feedback, including grounded knowledge of the environment, in a closed-loop interaction. Distinct from prior works, our approach augments user instructions into PDDL problems by leveraging both the abstract reasoning capabilities of LLMs and grounding mechanisms. By conducting various real-world long-horizon tasks, each consisting of seven distinct manipulatory skills, our results demonstrate that the IALP system can efficiently solve these tasks with an average success rate exceeding 80%. Our proposed method can operate as a high-level planner, equipping robots with substantial autonomy in unstructured environments through the utilization of multi-modal sensor inputs.

In the field of scene text spotting, previous OCR methods primarily relied on image encoders and pre-trained text information, but they often overlooked the advantages of incorporating human language instructions. To address this gap, we propose InstructOCR, an innovative instruction-based scene text spotting model that leverages human language instructions to enhance the understanding of text within images. Our framework employs both text and image encoders during training and inference, along with instructions meticulously designed based on text attributes. This approach enables the model to interpret text more accurately and flexibly. Extensive experiments demonstrate the effectiveness of our model and we achieve state-of-the-art results on widely used benchmarks. Furthermore, the proposed framework can be seamlessly applied to scene text VQA tasks. By leveraging instruction strategies during pre-training, the performance on downstream VQA tasks can be significantly improved, with a 2.6% increase on the TextVQA dataset and a 2.1% increase on the ST-VQA dataset. These experimental results provide insights into the benefits of incorporating human language instructions for OCR-related tasks.

Recent talking avatar generation models have made strides in achieving realistic and accurate lip synchronization with the audio, but often fall short in controlling and conveying detailed expressions and emotions of the avatar, making the generated video less vivid and controllable. In this paper, we propose a text-guided approach for generating emotionally expressive 2D avatars, offering fine-grained control, improved interactivity, and generalizability to the resulting video. Our framework, named InstructAvatar, leverages a natural language interface to control the emotion as well as the facial motion of avatars. Technically, we utilize GPT-4V to design an automatic annotation pipeline, constructing an instruction-video paired training dataset. This is combined with a novel two-branch diffusion-based generator to predict avatars using both audio and text instructions simultaneously. Experimental results demonstrate that InstructAvatar produces results that align well with both conditions, and outperforms existing methods in fine-grained emotion control, lip-sync quality, and naturalness.

Data augmentation is expected to bring about unseen features of training set, enhancing the model’s ability to generalize in situations where data is limited. Generative image models trained on large web-crawled datasets such as LAION are known to produce images with stereotypes and imperceptible bias when used to augment training data, owing to dataset misalignment and the generator’s ignorance of the downstream model. We improve downstream task awareness in generated images by proposing a task-aware fine-tuning strategy that actively detects failures of downstream task in the target model to fine-tune the generation process between epochs. The dynamic fine-tuning strategy is achieved by (1) inspecting misalignment between generated data and original data via VLM captioners and (2) adjusts both prompts and diffusion model so that the strategy dynamically guides the generator by focusing on the detected bias of VLM. This is done via re-captioning the overfitted data as well as finetuning the diffusion trajectory in a contrastive manner. To co-operate with the VLM captioner, the contrastive fine-tuning process dynamically adjusts different parts of the diffusion trajectory based on detected misalignment, thus shifting the the generated distribution away from making the downstream model overfit. Our experiments on few-shot class incremental learning show that our instruction-guided finetuning strategy consistently assists the downstream model with higher classification accuracy compared to generative data augmentation baselines such as Stable Diffusion and GPT-4o, and state-of-the-art non-generative strategies.

We present InstantSticker, a disentangled reconstruction pipeline based on Image-Based Lighting (IBL), which focuses on highly realistic decal blending, simulates stickers attached to the reconstructed surface, and allows for instant editing and real-time rendering. To achieve stereoscopic impression of the decal, we introduce shadow factor into IBL, which can be adaptively optimized during training. This allows the shadow brightness of surfaces to be accurately decomposed rather than baked into the diffuse color, ensuring that the edited texture exhibits authentic shading. To address the issues of warping and blurriness in previous methods, we apply As-Rigid-As-Possible (ARAP) parameterization to pre-unfold a specified area of the mesh and use the local UV mapping combined with a neural texture map to enhance the ability to express high-frequency details in that area. For instant editing, we utilize the Disney BRDF model, explicitly defining material colors with 3-channel diffuse albedo. This enables instant replacement of albedo RGB values during the editing process, avoiding the prolonged optimization required in previous approaches. In our experiment, we introduce the Ratio Variance Warping (RVW) metric to evaluate the local geometric warping of the decal area. Extensive experimental results demonstrate that our method surpasses previous decal blending methods in terms of editing quality, editing speed and rendering speed, achieving the state-of-the-art.

Text-conditioned image generation enables cross-modal comprehension. Recent emergence of many platforms have found applications in diverse domains like assisted designing and video gaming. However, there still exist challenges in existing platforms due to their expensive training and time-consuming generation processes. In this paper, we introduce an efficient text-conditioned image generation platform, termed InstantPainting. Unlike existing platforms based on large-scale pre-trained diffusion models, InstantPainting expands generative adversarial networks (GANs) to achieve efficient generation by using only about three percent pre-training data of other platforms. Compared to existing platforms, InstantPainting achieves the following functions at a very low deployment cost and approximately 4 to 5 times faster generation speeds: (1) Multi-category and multi-size image generation (2) Image stylization and controlled generation (3) Creative generation, including the generation of poetry pictures and counterfactual images. The proposed platform provides web application implementations for PC and mobile, users can create high-quality images directly through the user interface.

As artificial intelligence advances rapidly, particularly with the advent of GANs and diffusion models, the accuracy of Image Inpainting Localization (IIL) has become increasingly challenging. Current IIL methods face two main challenges: a tendency towards overconfidence, leading to incorrect predictions; and difficulty in detecting subtle tampering boundaries in inpainted images. In response, we propose a new paradigm that treats IIL as a conditional mask generation task utilizing diffusion models. Our method, InpDiffusion, utilizes the denoising process enhanced by the integration of image semantic conditions to progressively refine predictions. During denoising, we employ edge conditions and introduce a novel edge supervision strategy to enhance the model's perception of edge details in inpainted objects. Balancing the diffusion model's stochastic sampling with edge supervision of tampered image regions mitigates the risk of incorrect predictions from overconfidence and prevents the loss of subtle boundaries that can result from overly stochastic processes. Furthermore, we propose an innovative Dual-stream Multi-scale Feature Extractor (DMFE) for extracting multi-scale features, enhancing feature representation by considering both semantic and edge conditions of the inpainted images. Extensive experiments across challenging datasets demonstrate that the InpDiffusion significantly outperforms existing state-of-the-art methods in IIL tasks, while also showcasing excellent generalization capabilities and robustness.

Retinex-based methods have become a general approach for solving low-light image enhancement (LLIE). However, traditional methods require post-processing of illumination (e.g., gamma correction), which lacks adaptability and disrupts the illumination structure. Retinex-based deep networks typically follow a ‘decomposition-adjustment-exposure control’ process, which is redundant and lacks robustness. One major issue is the inaccuracy in estimating and decomposing the initial illumination. Accurate initial illumination can prevent further post-processing instability. We propose IniRetinex, rethinking the Retinex-based LLIE method from the perspective of initialization. By using neural networks to provide reasonable initial illumination and solving for smooth illumination through optimization, higher performance LLIE is achieved. We construct a two-layer convolutional neural network to capture the low-frequency structure of the image, adaptively compensating for classical initial illumination and avoiding additional post-processing. The network requires no pre-training and can be implemented in an unsupervised manner with just a few iterations, making it highly efficient. Additionally, we propose a new illumination optimization strategy by introducing an additional proximal penalty term, improving illumination in areas with varying levels and enhancing image details. Extensive experiments on various low-light image datasets demonstrate that our method achieves state-of-the-art (SOTA) results on multiple benchmarks, offering higher stability and inference efficiency compared to current advanced methods.

In practical applications, it is often necessary to transfer knowledge from large pretrained models to small ones with various architectures for tackling different tasks. The Learngene framework, proposed recently, firstly extracts one compact module termed as learngene from a large well-trained model, after which learngene is used to build descendant models for handling diverse tasks. In this paper, we aim to explore extracting and inheriting learngene which can be generalized across different model architectures and tasks, remaining understudied in previous works. Inspired by the existing observations that large kernel convolutional neural networks (CNNs) exhibit significant generalization potential across various architectures and tasks, we propose a novel two-stage Learngene method termed CLKG (Convolutional Learngene for Knowledge Generalization), which inherits convolutional kernels containing generalized knowledge as learngene to build diverse models for multiple tasks. Specifically, we construct an auxiliary model comprised of small kernels and train it through dense feature distillation to inherit the feature extraction ability from large kernel CNNs. After distillation, we select certain kernels from the auxiliary model as learngene based on three criteria: direct kernel extraction, priority to edge kernels, and continuous kernel selection. Subsequently, we adapt learngene according to the width of the descendant models and use it to initialize the backbone part of descendant models. Experiments on diverse vision tasks such as image classification, object detection and semantic segmentation demonstrate the superiority of CLKG. For example, compared with from scratch training, it brings 2.89% improvements on VOC12+SBD, and reduces around 2x training data volume and training epochs to achieve better results. Furthermore, compared to knowledge distillation method, CLKG significantly reduces negative transfer on certain datasets, e.g., achieves 1.88% performance improvements on NAO dataset despite domain differences.