IJCAI 2024 论文列表

Recently, deepfake detection research has been actively conducted. While many deepfake detectors have been proposed, validating the practicality of such systems against real world settings has not been explored much. Indeed, there are some gaps and disparities when they are applied in the real world. In this work, we developed a real time integrated web-based deepfake detection system, iFakeDetector, which incorporates the recent high performing deepfake detectors, and enables easy access for non-expert users to evaluate deepfake videos. Our system takes a deepfake video as input, allowing users to upload videos and select different detectors, and provides detection results on whether the uploaded video is a deepfake or not. Also, we provide an analysis tool that enables the video to be analyzed on a frame-by-frame basis with the probability of each frame being manipulated. Finally, we tested and deployed iFakeDetector in a real world scenario to verify its practicality and feasibility.

Sequential recommendation models predict the next item in a sequence of user-item interactions, akin to how language models predict the next tokens. These models often adapt language model architectures, treating item IDs as if they were token IDs. However, the number of potential items in recommender systems makes calculating the interaction probability for all items impractical during training; therefore, recommender systems frequently employ negative sampling, where the model learns to differentiate between actual user interactions (positives) and randomly chosen non-interactions (negatives), often using Binary Cross-Entropy (BCE) the loss, framing the problem as a binary classification task. We demonstrate that using negative sampling with BCE can lead to overconfidence, where the model's predicted probabilities for user interactions are higher than the actual probabilities. Although the actual score magnitude is not important for ranking items (only the order of scores matters), overconfidence leads to training instability when using Binary Cross-Entropy (BCE) loss. We show that overconfidence explains the performance gap between two leading sequential recommendation models, SASRec and BERT4Rec -- the former uses negative sampling, while the latter does not. To counter overconfidence, we introduce Generalised Binary Cross-Entropy (gBCE) loss and the gSASRec model that utilises gBCE. We mathematically prove and empirically validate that gSASRec effectively addresses the issue of overconfidence. Consequently, gSASRec's effectiveness is better than that of SASRec and matches the state of the BERT4Rec while retaining negative sampling. On the Gowalla dataset with more than 1MM items, where training BERT4Rec is infeasible, gSASRec outperforms the original SASRec model by 41% in terms of NDCG@10.

Any-angle path planning is an extension of traditional path-planning algorithms that aims to generate smoother and shorter paths in graphs by allowing any-angle moves between vertices, rather than being restricted by edges. Many any-angle path-planning algorithms have been proposed, such as Theta*, Block A* and Anya, but most of them are designed only for static environments, which is not applicable when dynamic obstacles are present. Time-Optimal Any-Angle Safe-Interval Path Planning (TO-AA-SIPP) was developed to fill this gap, which can find an optimal collision-free any-angle path that minimizes the traversal time. However, as indicated by its authors, TO-AA-SIPP may not be efficient enough to be used in multi-agent pathfinding (MAPF). Therefore, this paper presents a new algorithm Zeta*-SIPP to improve TO-AA-SIPP by means of 1) reducing useless search nodes that have no contribution to finding optimal solutions, and 2) introducing Field of View (FoV) instead of Line of Sight (LoS) to speed up visibility checks with static obstacles. Benchmark experiments showed that Zeta*-SIPP reduced the computation time of TO-AA-SIPP by around 70%-90% on average.

Drug-drug interactions (DDIs) can result in various pharmacological changes, which can be categorized into different classes known as DDI events (DDIEs). In recent years, previously unobserved/unseen DDIEs have been emerging, posing a new classification task when unseen classes have no labelled instances in the training stage, which is formulated as a zero-shot DDIE prediction (ZS-DDIE) task. However, existing computational methods are not directly applicable to ZS-DDIE, which has two primary challenges: obtaining suitable DDIE representations and handling the class imbalance issue. To overcome these challenges, we propose a novel method named ZeroDDI for the ZS-DDIE task. Specifically, we design a biological semantic enhanced DDIE representation learning module, which emphasizes the key biological semantics and distills discriminative molecular substructure-related semantics for DDIE representation learning. Furthermore, we propose a dual-modal uniform alignment strategy to distribute drug pair representations and DDIE semantic representations uniformly in unit sphere and align the matched ones, which can mitigate the issue of class imbalance. Extensive experiments showed that ZeroDDI surpasses the baselines and indicate that it is a promising tool for detecting unseen DDIEs. Our code has been released in https://github.com/wzy-Sarah/ZeroDDI.

Conventional deep learning methods typically employ supervised learning for drug response prediction (DRP). This entails dependence on labeled response data from drugs for model training. However, practical applications in the preclinical drug screening phase demand that DRP models predict responses for novel compounds, often with unknown drug responses. This presents a challenge, rendering supervised deep learning methods unsuitable for such scenarios. In this paper, we propose a zero-shot learning solution for the DRP task in preclinical drug screening. Specifically, we propose a Multi-branch Multi-Source Domain Adaptation Test Enhancement Plug-in, called MSDA. MSDA can be seamlessly integrated with conventional DRP methods, learning invariant features from the prior response data of similar drugs to enhance real-time predictions of unlabeled compounds. The results of experiments on two large drug response datasets showed that MSDA efficiently predicts drug responses for novel compounds, leading to a general performance improvement of 5-10% in the preclinical drug screening phase. The significance of this solution resides in its potential to accelerate the drug discovery process, improve drug candidate assessment, and facilitate the success of drug discovery. The code is available at https://github.com/DrugD/MSDA.

Image-to-video (I2V) generation aims to create a video sequence from a single image, which requires high temporal coherence and visual fidelity. However, existing approaches suffer from inconsistency of character appearances and poor preservation of fine details. Moreover, they require a large amount of video data for training, which can be computationally demanding. To address these limitations, we propose PoseAnimate, a novel zero-shot I2V framework for character animation. PoseAnimate contains three key components: 1) a Pose-Aware Control Module (PACM) that incorporates diverse pose signals into text embeddings, to preserve character-independent content and maintain precise alignment of actions. 2) a Dual Consistency Attention Module (DCAM) that enhances temporal consistency and retains character identity and intricate background details. 3) a Mask-Guided Decoupling Module (MGDM) that refines distinct feature perception abilities, improving animation fidelity by decoupling the character and background. We also propose a Pose Alignment Transition Algorithm (PATA) to ensure smooth action transition. Extensive experiment results demonstrate that our approach outperforms the state-of-the-art training-based methods in terms of character consistency and detail fidelity. Moreover, it maintains a high level of temporal coherence throughout the generated animations.

Zero-shot sketch-based image retrieval (ZS-SBIR), aiming to recognize and retrieve relevant photos based on freehand sketch queries that belong to unseen categories in the search set, has sparked considerable interest, benefiting from the rapid advancements in multimodal learning and feature representation research. Despite the recent improvements in performance, there are still rooms for refining feature representation and thus enhancing the generalization capabilities of the models. Most of the existing research efforts have primarily focused on learning the feature distribution of modalities within specific datasets, without considering the broader dataset-agnostic `population distribution' of relevant modalities. In this paper, we investigate the modality population distribution and apply such knowledge to guide feature learning. Specifically, we propose a modality capacity constraint loss to control the learning of population distribution for sketches and photos. This loss can be effectively combined with retrieval loss (e.g., triplet loss) or classification loss (e.g., InfoNCE loss) to enhance the performance of ZS-SBIR, through the fine-tuning process of pre-trained models like CLIP and DINO. Extensive experiment results have demonstrated our significant performance improvements, achieving an increase of 7.3%/3.2% and 19.9%/10.3% in terms of mAP@200/P@200 compared to the state-of-the-art models on CLIP and DINO, respectively, on the Sketchy-ext dataset (split 2). Data, code, and supplementary information are available at https://github.com/YHdian0716/ZS-SBIR-MCC.git.

This research introduces XGA-Osteo, an innovative approach that leverages Explainable Artificial Intelligence (XAI) to enhance the accuracy and interpretability of knee osteoarthritis diagnosis. Recent studies have utilized AI approaches to automate the diagnosis using knee joint X-ray images. However, these studies have primarily focused on predicting the severity of osteoarthritis without providing additional information to assist doctors in their diagnoses. In addition to accurately diagnosing the severity of the condition, XGA-Osteo generates an anomaly map, produced from a reconstructed image of a healthy knee using adversarial learning. Thus, the abnormal regions in X-ray images can be highlighted, offering valuable supplementary information to medical experts during the diagnosis process.

Explaining the singability of lyrics is an important but missing ability of language models (LMs) in song lyrics generation. This ability allows songwriters to quickly assess if LM-generated lyrics can be sung harmoniously with melodies and helps singers align lyrics with melodies during practice. This paper presents XAI-Lyricist, leveraging musical prosody to guide LMs in generating singable lyrics and providing human-understandable singability explanations. We employ a Transformer model to generate lyrics under musical prosody constraints and provide demonstrations of the lyrics' prosody patterns as singability explanations. XAI-Lyricist is evaluated by computational metrics (perplexity, prosody-BLEU) and a human-grounded study (human ratings, average time and number of attempts for singing). Experimental results show that musical prosody can significantly improve the singability of LM-generated lyrics. A controlled study with 14 singers also confirms the usefulness of the provided explanations in helping them to interpret lyrical singability faster than reading plain text lyrics.

Transformer-based LLMs are becoming increasingly important in various AI applications. However, apart from the success of LLMs, the explosive demand of long context handling capabilities is a key and in-time problem for both academia and industry. Due to the limitations from the quadratic complexity of the attention mechanism, long context scenarios require much more resources for LLM development and deployment, bringing huge challenges to the underlying AI infrastructure. Meanwhile, we observe that there is a trend of reviving previous efficient attention mechanisms to latest LLMs. However, it still remains an open question about how to select from these diverse approaches in practice. In this paper, we answer this question from several aspects. First, we revisit these latest long-context LLM innovations and discuss their relationship with prior approaches with a novel and comprehensive taxonomy. Next, we conduct a thorough evaluation over various types of workloads considering both efficiency and effectiveness. Finally, we provide an in-depth analysis, summarize our key findings, and offer insightful suggestions on the trade-offs of designing and deploying efficient attention mechanisms for Transformer-based LLMs.

The effectiveness of traffic light control has been significantly improved by current reinforcement learning-based approaches via better cooperation among multiple traffic lights. However, a persisting issue remains: how to obtain a multi-agent traffic signal control algorithm with remarkable transferability across diverse cities? In this paper, we propose a Transformer on Transformer (TonT) model for cross-city meta multi-agent traffic signal control, named as X-Light: We input the full Markov Decision Process trajectories, and the Lower Transformer aggregates the states, actions, rewards among the target intersection and its neighbors within a city, and the Upper Transformer learns the general decision trajectories across different cities. This dual-level approach bolsters the model's robust generalization and transferability. Notably, when directly transferring to unseen scenarios, ours surpasses all baseline methods with +7.91% on average, and even +16.3% in some cases, yielding the best results.

Few-shot learning (FSL) aims to recognize new categories given limited training samples. The core challenge is to avoid overfitting to the minimal data while ensuring good generalization to novel classes. One mainstream method employs prototypes from visual feature extractors as classifier weight and the performance depends on the quality of the prototype. Since different categories may have similar visual features, the visual prototype has limitations. This is because existing methods only learn a simple visual feature extractor during the pre-training stage but neglect the importance of a well-developed feature space for the prototype. We introduce the Semantic Enhanced Visual Prototype framework (SEVpro) to address this issue. SEVpro refines prototype learning from the pre-training stage and serves as a versatile plug-and-play framework for all prototype-based FSL methods. Specifically, we enhance prototype discriminability by transforming semantic embeddings into the visual space, aiding in separating categories with similar visual features. For novel class learning, we leverage knowledge from base classes and incorporate semantic information to elevate prototype quality further. Meanwhile, extensive experiments on FSL benchmarks and ablation studies demonstrate the superiority of our proposed SEVpro for FSL.

Current Vision-and-Language Navigation (VLN) tasks mainly employ textual instructions to guide agents. However, being inherently abstract, the same textual instruction can be associated with different visual signals, causing severe ambiguity and limiting the transfer of prior knowledge in the vision domain from the user to the agent. To fill this gap, we propose Vision-and-Language Navigation with Multi-modal Prompts (VLN-MP), a novel task augmenting traditional VLN by integrating both natural language and images in instructions. VLN-MP not only maintains backward compatibility by effectively handling text-only prompts but also consistently shows advantages with different quantities and relevance of visual prompts. Possible forms of visual prompts include both exact and similar object images, providing adaptability and versatility in diverse navigation scenarios. To evaluate VLN-MP under a unified framework, we implement a new benchmark that offers: (1) a training-free pipeline to transform textual instructions into multi-modal forms with landmark images; (2) diverse datasets with multi-modal instructions for different downstream tasks; (3) a novel module designed to process various image prompts for seamless integration with state-of-the-art VLN models. Extensive experiments on four VLN benchmarks (R2R, RxR, REVERIE, CVDN) show that incorporating visual prompts would significantly boost navigation performance. While maintaining efficiency with text-only prompts, VLN-MP enables agents to navigate in the pre-explore setting and outperform text-based models, showing its broader applicability. Code is available at https://github.com/honghd16/VLN-MP.

Image harmonization, aiming to seamlessly blend extraneous foreground objects with background images, is a promising and challenging task.Ensuring a synthetic image appears realistic requires maintaining consistency in visual characteristics, such as texture and style, across global and semantic regions.In this paper, We approach image harmonization as a semantic routed style transfer problem, and propose an imageharmonization model by routing semantic similarity explicitly to enhance the consistency of appearance characteristics.To refine calculate the similarity between the composed foreground and background instance, we propose an InstanceSimilarity Evaluation Module(ISEM).To harness analogous semantic information effectively, we further introduceStyle Transfer Block(STB) to establish fine-grained foreground-background semantic correlation.Our method has achieved excellent experimental results on existing datasets and our model outperforms the state-of-the-art by a margin of 0.45 dB on iHarmony4 dataset.

Graph masked autoencoders (GMAE) have emerged as a significant advancement in self-supervised pre-training for graph-structured data. Previous GMAE models primarily utilize a straightforward random masking strategy for nodes or edges during training. However, this strategy fails to consider the varying significance of different nodes within the graph structure. In this paper, we investigate the potential of leveraging the graph's structural composition as a fundamental and unique prior in the masked pre-training process. To this end, we introduce a novel structure-guided masking strategy (i.e., StructMAE), designed to refine the existing GMAE models. StructMAE involves two steps: 1) Structure-based Scoring: Each node is evaluated and assigned a score reflecting its structural significance. Two distinct types of scoring manners are proposed: predefined and learnable scoring. 2) Structure-guided Masking: With the obtained assessment scores, we develop an easy-to-hard masking strategy that gradually increases the structural awareness of the self-supervised reconstruction task. Specifically, the strategy begins with random masking and progresses to masking structure-informative nodes based on the assessment scores. This design gradually and effectively guides the model in learning graph structural information. Furthermore, extensive experiments consistently demonstrate that our StructMAE method outperforms existing state-of-the-art GMAE models in both unsupervised and transfer learning tasks. Codes are available at https: //github.com/LiuChuang0059/StructMAE.

Multi-modality image fusion aims to integrate images from multiple sensors, producing an image that is visually appealing and offers more comprehensive information than any single one. To ensure high visual quality and facilitate accurate subsequent perception tasks, previous methods have often cascaded networks using weighted loss functions. However, such simplistic strategies struggle to truly achieve the "Best of Both Worlds", and the adjustment of numerous hand-crafted parameters becomes burdensome. To address these challenges, this paper introduces a Compact, Automatic and Flexible framework, dubbed CAF, designed for infrared and visible image fusion, along with subsequent tasks. Concretely, we recast the combined problem of fusion and perception into a single objective, allowing mutual optimization of information from both tasks. Then we also utilize the perception task to inform the design of fusion loss functions, facilitating the automatic identification of optimal fusion objectives tailored to the task. Furthermore, CAF can support seamless integration with existing approaches easily, offering flexibility in adapting to various tasks and network structures. Extensive experiments demonstrate the superiority of CAF, which not only produces visually admirable fused results but also realizes 1.7 higher detection mAP@.5 and 2.0 higher segmentation mIoU than the state-of-the-art methods. The code is available at https://github.com/RollingPlain/CAF_IVIF.

While in-processing fairness approaches show promise in mitigating bias predictions, their potential impact on privacy leakage remains under-explored. We aim to address this gap by assessing the privacy risks of fairness-enhanced binary classifiers with membership inference attacks (MIAs). Surprisingly, our results reveal that these fairness interventions exhibit increased resilience against existing attacks, indicating that enhancing fairness does not necessarily lead to privacy compromises. However, we find current attack methods are ineffective as they typically degrade into simple threshold models with limited attack effectiveness. Following this observation, we discover a novel threat dubbed Fairness Discrepancy Membership Inference Attacks (FD-MIA) that exploits prediction discrepancies between fair and biased models. This attack reveals more potent vulnerabilities and poses significant privacy risks to model privacy. Extensive experiments across multiple datasets, attack methods, and representative fairness approaches confirm our findings and demonstrate the efficacy of the proposed attack method. Our study exposes the overlooked privacy threats in fairness studies, advocating for thorough evaluations of potential security vulnerabilities before model deployments.

Compositionality is thought to be a key component of language, and various compositional benchmarks have been developed to empirically probe the compositional generalization of existing sequence processing models. These benchmarks often highlight failures of existing models, but it is not clear why these models fail in this way. In this paper, we seek to theoretically understand the role the compositional structure of the models plays in these failures and how this structure relates to their expressivity and sample complexity. We propose a general neuro-symbolic definition of compositional functions and their compositional complexity. We then show how various existing general and special purpose sequence processing models (such as recurrent, convolution and attention-based ones) fit this definition and use it to analyze their compositional complexity. Finally, we provide theoretical guarantees for the expressivity and systematic generalization of compositional models that explicitly depend on our proposed definition and highlighting factors which drive poor empirical performance.

We study letter grading schemes, which are routinely employed for evaluating student performance. Typically, a numerical score obtained via one or more evaluations is converted into a letter grade (e.g., A+, B-, etc.) by associating a disjoint interval of numerical scores to each letter grade. We propose the first model for studying the (de)motivational effects of such grading on the students and, consequently, on their performance in future evaluations. We use the model to compare uniform letter grading schemes, in which the range of scores is divided into equal-length parts that are mapped to the letter grades, to numerical scoring, in which the score is not converted to any letter grade (equivalently, every score is its own letter grade). Theoretically, we identify realistic conditions under which numerical scoring is better than any uniform letter grading scheme. Our experiments confirm that this holds under even weaker conditions, but also find cases where the converse occurs.