Abstract:Due to the common content of anatomy, radiology images with their corresponding reports exhibit high similarity. Such inherent data bias can predispose automatic report generation models to learn entangled and spurious representations resulting in misdiagnostic reports. To tackle these, we propose a novel \textbf{Co}unter\textbf{F}actual \textbf{E}xplanations-based framework (CoFE) for radiology report generation. Counterfactual explanations serve as a potent tool for understanding how decisions made by algorithms can be changed by asking ``what if'' scenarios. By leveraging this concept, CoFE can learn non-spurious visual representations by contrasting the representations between factual and counterfactual images. Specifically, we derive counterfactual images by swapping a patch between positive and negative samples until a predicted diagnosis shift occurs. Here, positive and negative samples are the most semantically similar but have different diagnosis labels. Additionally, CoFE employs a learnable prompt to efficiently fine-tune the pre-trained large language model, encapsulating both factual and counterfactual content to provide a more generalizable prompt representation. Extensive experiments on two benchmarks demonstrate that leveraging the counterfactual explanations enables CoFE to generate semantically coherent and factually complete reports and outperform in terms of language generation and clinical efficacy metrics.
Abstract:In learning vision-language representations from web-scale data, the contrastive language-image pre-training (CLIP) mechanism has demonstrated a remarkable performance in many vision tasks. However, its application to the widely studied video quality assessment (VQA) task is still an open issue. In this paper, we propose an efficient and effective CLIP-based Transformer method for the VQA problem (CLIPVQA). Specifically, we first design an effective video frame perception paradigm with the goal of extracting the rich spatiotemporal quality and content information among video frames. Then, the spatiotemporal quality features are adequately integrated together using a self-attention mechanism to yield video-level quality representation. To utilize the quality language descriptions of videos for supervision, we develop a CLIP-based encoder for language embedding, which is then fully aggregated with the generated content information via a cross-attention module for producing video-language representation. Finally, the video-level quality and video-language representations are fused together for final video quality prediction, where a vectorized regression loss is employed for efficient end-to-end optimization. Comprehensive experiments are conducted on eight in-the-wild video datasets with diverse resolutions to evaluate the performance of CLIPVQA. The experimental results show that the proposed CLIPVQA achieves new state-of-the-art VQA performance and up to 37% better generalizability than existing benchmark VQA methods. A series of ablation studies are also performed to validate the effectiveness of each module in CLIPVQA.
Abstract:The few-shot fine-tuning of Latent Diffusion Models (LDMs) has enabled them to grasp new concepts from a limited number of images. However, given the vast amount of personal images accessible online, this capability raises critical concerns about civil privacy. While several previous defense methods have been developed to prevent such misuse of LDMs, they typically assume that the textual prompts used by data protectors exactly match those employed by data exploiters. In this paper, we first empirically demonstrate that breaking this assumption, i.e., in cases where discrepancies exist between the textual conditions used by protectors and exploiters, could substantially reduce the effectiveness of these defenses. Furthermore, considering the visual encoder's independence from textual prompts, we delve into the visual encoder and thoroughly investigate how manipulating the visual encoder affects the few-shot fine-tuning process of LDMs. Drawing on these insights, we propose a simple yet effective method called \textbf{Prompt-Independent Defense (PID)} to safeguard privacy against LDMs. We show that PID can act as a strong privacy shield on its own while requiring significantly less computational power. We believe our studies, along with the comprehensive understanding and new defense method, provide a notable advance toward reliable data protection against LDMs.
Abstract:Knowledge distillation (KD) is a widely adopted and effective method for compressing models in object detection tasks. Particularly, feature-based distillation methods have shown remarkable performance. Existing approaches often ignore the uncertainty in the teacher model's knowledge, which stems from data noise and imperfect training. This limits the student model's ability to learn latent knowledge, as it may overly rely on the teacher's imperfect guidance. In this paper, we propose a novel feature-based distillation paradigm with knowledge uncertainty for object detection, termed "Uncertainty Estimation-Discriminative Knowledge Extraction-Knowledge Transfer (UET)", which can seamlessly integrate with existing distillation methods. By leveraging the Monte Carlo dropout technique, we introduce knowledge uncertainty into the training process of the student model, facilitating deeper exploration of latent knowledge. Our method performs effectively during the KD process without requiring intricate structures or extensive computational resources. Extensive experiments validate the effectiveness of our proposed approach across various distillation strategies, detectors, and backbone architectures. Specifically, following our proposed paradigm, the existing FGD method achieves state-of-the-art (SoTA) performance, with ResNet50-based GFL achieving 44.1% mAP on the COCO dataset, surpassing the baselines by 3.9%.
Abstract:Predicting genetic mutations from whole slide images is indispensable for cancer diagnosis. However, existing work training multiple binary classification models faces two challenges: (a) Training multiple binary classifiers is inefficient and would inevitably lead to a class imbalance problem. (b) The biological relationships among genes are overlooked, which limits the prediction performance. To tackle these challenges, we innovatively design a Biological-knowledge enhanced PathGenomic multi-label Transformer to improve genetic mutation prediction performances. BPGT first establishes a novel gene encoder that constructs gene priors by two carefully designed modules: (a) A gene graph whose node features are the genes' linguistic descriptions and the cancer phenotype, with edges modeled by genes' pathway associations and mutation consistencies. (b) A knowledge association module that fuses linguistic and biomedical knowledge into gene priors by transformer-based graph representation learning, capturing the intrinsic relationships between different genes' mutations. BPGT then designs a label decoder that finally performs genetic mutation prediction by two tailored modules: (a) A modality fusion module that firstly fuses the gene priors with critical regions in WSIs and obtains gene-wise mutation logits. (b) A comparative multi-label loss that emphasizes the inherent comparisons among mutation status to enhance the discrimination capabilities. Sufficient experiments on The Cancer Genome Atlas benchmark demonstrate that BPGT outperforms the state-of-the-art.
Abstract:The goal of No-Reference Image Quality Assessment (NR-IQA) is to predict the perceptual quality of an image in line with its subjective evaluation. To put the NR-IQA models into practice, it is essential to study their potential loopholes for model refinement. This paper makes the first attempt to explore the black-box adversarial attacks on NR-IQA models. Specifically, we first formulate the attack problem as maximizing the deviation between the estimated quality scores of original and perturbed images, while restricting the perturbed image distortions for visual quality preservation. Under such formulation, we then design a Bi-directional loss function to mislead the estimated quality scores of adversarial examples towards an opposite direction with maximum deviation. On this basis, we finally develop an efficient and effective black-box attack method against NR-IQA models. Extensive experiments reveal that all the evaluated NR-IQA models are vulnerable to the proposed attack method. And the generated perturbations are not transferable, enabling them to serve the investigation of specialities of disparate IQA models.
Abstract:Information extraction techniques, including named entity recognition (NER) and relation extraction (RE), are crucial in many domains to support making sense of vast amounts of unstructured text data by identifying and connecting relevant information. Such techniques can assist researchers in extracting valuable insights. In this paper, we introduce the Entity-aware Masking for Biomedical Relation Extraction (EMBRE) method for biomedical relation extraction, as applied in the context of the BioRED challenge Task 1, in which human-annotated entities are provided as input. Specifically, we integrate entity knowledge into a deep neural network by pretraining the backbone model with an entity masking objective. We randomly mask named entities for each instance and let the model identify the masked entity along with its type. In this way, the model is capable of learning more specific knowledge and more robust representations. Then, we utilize the pre-trained model as our backbone to encode language representations and feed these representations into two multilayer perceptron (MLPs) to predict the logits for relation and novelty, respectively. The experimental results demonstrate that our proposed method can improve the performances of entity pair, relation and novelty extraction over our baseline.
Abstract:Video Semantic Segmentation (VSS) involves assigning a semantic label to each pixel in a video sequence. Prior work in this field has demonstrated promising results by extending image semantic segmentation models to exploit temporal relationships across video frames; however, these approaches often incur significant computational costs. In this paper, we propose an efficient mask propagation framework for VSS, called MPVSS. Our approach first employs a strong query-based image segmentor on sparse key frames to generate accurate binary masks and class predictions. We then design a flow estimation module utilizing the learned queries to generate a set of segment-aware flow maps, each associated with a mask prediction from the key frame. Finally, the mask-flow pairs are warped to serve as the mask predictions for the non-key frames. By reusing predictions from key frames, we circumvent the need to process a large volume of video frames individually with resource-intensive segmentors, alleviating temporal redundancy and significantly reducing computational costs. Extensive experiments on VSPW and Cityscapes demonstrate that our mask propagation framework achieves SOTA accuracy and efficiency trade-offs. For instance, our best model with Swin-L backbone outperforms the SOTA MRCFA using MiT-B5 by 4.0% mIoU, requiring only 26% FLOPs on the VSPW dataset. Moreover, our framework reduces up to 4x FLOPs compared to the per-frame Mask2Former baseline with only up to 2% mIoU degradation on the Cityscapes validation set. Code is available at https://github.com/ziplab/MPVSS.
Abstract:The AI model has surpassed human players in the game of Go, and it is widely believed that the AI model has encoded new knowledge about the Go game beyond human players. In this way, explaining the knowledge encoded by the AI model and using it to teach human players represent a promising-yet-challenging issue in explainable AI. To this end, mathematical supports are required to ensure that human players can learn accurate and verifiable knowledge, rather than specious intuitive analysis. Thus, in this paper, we extract interaction primitives between stones encoded by the value network for the Go game, so as to enable people to learn from the value network. Experiments show the effectiveness of our method.
Abstract:Medical artificial general intelligence (MAGI) enables one foundation model to solve different medical tasks, which is very practical in the medical domain. It can significantly reduce the requirement of large amounts of task-specific data by sufficiently sharing medical knowledge among different tasks. However, due to the challenges of designing strongly generalizable models with limited and complex medical data, most existing approaches tend to develop task-specific models. To take a step towards MAGI, we propose a new paradigm called Medical-knOwledge-enhanced mulTimOdal pretRaining (MOTOR). In MOTOR, we combine two kinds of basic medical knowledge, i.e., general and specific knowledge, in a complementary manner to boost the general pretraining process. As a result, the foundation model with comprehensive basic knowledge can learn compact representations from pretraining radiographic data for better cross-modal alignment. MOTOR unifies the understanding and generation, which are two kinds of core intelligence of an AI system, into a single medical foundation model, to flexibly handle more diverse medical tasks. To enable a comprehensive evaluation and facilitate further research, we construct a medical multimodal benchmark including a wide range of downstream tasks, such as chest x-ray report generation and medical visual question answering. Extensive experiments on our benchmark show that MOTOR obtains promising results through simple task-oriented adaptation. The visualization shows that the injected knowledge successfully highlights key information in the medical data, demonstrating the excellent interpretability of MOTOR. Our MOTOR successfully mimics the human practice of fulfilling a "medical student" to accelerate the process of becoming a "specialist". We believe that our work makes a significant stride in realizing MAGI.