SafeCFG: Redirecting Harmful Classifier-Free Guidance for Safe Generation

Diffusion models (DMs) have demonstrated exceptional performance in text-to-image (T2I) tasks, leading to their widespread use. With the introduction of classifier-free guidance (CFG), the quality of images generated by DMs is improved. However, DMs can generate more harmful images by maliciously guiding the image generation process through CFG. Some safe guidance methods aim to mitigate the risk of generating harmful images but often reduce the quality of clean image generation. To address this issue, we introduce the Harmful Guidance Redirector (HGR), which redirects harmful CFG direction while preserving clean CFG direction during image generation, transforming CFG into SafeCFG and achieving high safety and quality generation. We train HGR to redirect multiple harmful CFG directions simultaneously, demonstrating its ability to eliminate various harmful elements while preserving high-quality generation. Additionally, we find that HGR can detect image harmfulness, allowing for unsupervised fine-tuning of safe diffusion models without pre-defined clean or harmful labels. Experimental results show that by incorporating HGR, images generated by diffusion models achieve both high quality and strong safety, and safe DMs trained through unsupervised methods according to the harmfulness detected by HGR also exhibit good safety performance. The codes will be publicly available.

Query-centric Audio-Visual Cognition Network for Moment Retrieval, Segmentation and Step-Captioning

Video has emerged as a favored multimedia format on the internet. To better gain video contents, a new topic HIREST is presented, including video retrieval, moment retrieval, moment segmentation, and step-captioning. The pioneering work chooses the pre-trained CLIP-based model for video retrieval, and leverages it as a feature extractor for other three challenging tasks solved in a multi-task learning paradigm. Nevertheless, this work struggles to learn the comprehensive cognition of user-preferred content, due to disregarding the hierarchies and association relations across modalities. In this paper, guided by the shallow-to-deep principle, we propose a query-centric audio-visual cognition (QUAG) network to construct a reliable multi-modal representation for moment retrieval, segmentation and step-captioning. Specifically, we first design the modality-synergistic perception to obtain rich audio-visual content, by modeling global contrastive alignment and local fine-grained interaction between visual and audio modalities. Then, we devise the query-centric cognition that uses the deep-level query to perform the temporal-channel filtration on the shallow-level audio-visual representation. This can cognize user-preferred content and thus attain a query-centric audio-visual representation for three tasks. Extensive experiments show QUAG achieves the SOTA results on HIREST. Further, we test QUAG on the query-based video summarization task and verify its good generalization.

SSE-SAM: Balancing Head and Tail Classes Gradually through Stage-Wise SAM

Real-world datasets often exhibit a long-tailed distribution, where vast majority of classes known as tail classes have only few samples. Traditional methods tend to overfit on these tail classes. Recently, a new approach called Imbalanced SAM (ImbSAM) is proposed to leverage the generalization benefits of Sharpness-Aware Minimization (SAM) for long-tailed distributions. The main strategy is to merely enhance the smoothness of the loss function for tail classes. However, we argue that improving generalization in long-tail scenarios requires a careful balance between head and tail classes. We show that neither SAM nor ImbSAM alone can fully achieve this balance. For SAM, we prove that although it enhances the model's generalization ability by escaping saddle point in the overall loss landscape, it does not effectively address this for tail-class losses. Conversely, while ImbSAM is more effective at avoiding saddle points in tail classes, the head classes are trained insufficiently, resulting in significant performance drops. Based on these insights, we propose Stage-wise Saddle Escaping SAM (SSE-SAM), which uses complementary strengths of ImbSAM and SAM in a phased approach. Initially, SSE-SAM follows the majority sample to avoid saddle points of the head-class loss. During the later phase, it focuses on tail-classes to help them escape saddle points. Our experiments confirm that SSE-SAM has better ability in escaping saddles both on head and tail classes, and shows performance improvements.

Bidirectional Logits Tree: Pursuing Granularity Reconcilement in Fine-Grained Classification

This paper addresses the challenge of Granularity Competition in fine-grained classification tasks, which arises due to the semantic gap between multi-granularity labels. Existing approaches typically develop independent hierarchy-aware models based on shared features extracted from a common base encoder. However, because coarse-grained levels are inherently easier to learn than finer ones, the base encoder tends to prioritize coarse feature abstractions, which impedes the learning of fine-grained features. To overcome this challenge, we propose a novel framework called the Bidirectional Logits Tree (BiLT) for Granularity Reconcilement. The key idea is to develop classifiers sequentially from the finest to the coarsest granularities, rather than parallelly constructing a set of classifiers based on the same input features. In this setup, the outputs of finer-grained classifiers serve as inputs for coarser-grained ones, facilitating the flow of hierarchical semantic information across different granularities. On top of this, we further introduce an Adaptive Intra-Granularity Difference Learning (AIGDL) approach to uncover subtle semantic differences between classes within the same granularity. Extensive experiments demonstrate the effectiveness of our proposed method.

EmoDubber: Towards High Quality and Emotion Controllable Movie Dubbing

Given a piece of text, a video clip, and a reference audio, the movie dubbing task aims to generate speech that aligns with the video while cloning the desired voice. The existing methods have two primary deficiencies: (1) They struggle to simultaneously hold audio-visual sync and achieve clear pronunciation; (2) They lack the capacity to express user-defined emotions. To address these problems, we propose EmoDubber, an emotion-controllable dubbing architecture that allows users to specify emotion type and emotional intensity while satisfying high-quality lip sync and pronunciation. Specifically, we first design Lip-related Prosody Aligning (LPA), which focuses on learning the inherent consistency between lip motion and prosody variation by duration level contrastive learning to incorporate reasonable alignment. Then, we design Pronunciation Enhancing (PE) strategy to fuse the video-level phoneme sequences by efficient conformer to improve speech intelligibility. Next, the speaker identity adapting module aims to decode acoustics prior and inject the speaker style embedding. After that, the proposed Flow-based User Emotion Controlling (FUEC) is used to synthesize waveform by flow matching prediction network conditioned on acoustics prior. In this process, the FUEC determines the gradient direction and guidance scale based on the user's emotion instructions by the positive and negative guidance mechanism, which focuses on amplifying the desired emotion while suppressing others. Extensive experimental results on three benchmark datasets demonstrate favorable performance compared to several state-of-the-art methods.

Expanding Sparse Tuning for Low Memory Usage

Parameter-efficient fine-tuning (PEFT) is an effective method for adapting pre-trained vision models to downstream tasks by tuning a small subset of parameters. Among PEFT methods, sparse tuning achieves superior performance by only adjusting the weights most relevant to downstream tasks, rather than densely tuning the whole weight matrix. However, this performance improvement has been accompanied by increases in memory usage, which stems from two factors, i.e., the storage of the whole weight matrix as learnable parameters in the optimizer and the additional storage of tunable weight indexes. In this paper, we propose a method named SNELL (Sparse tuning with kerNELized LoRA) for sparse tuning with low memory usage. To achieve low memory usage, SNELL decomposes the tunable matrix for sparsification into two learnable low-rank matrices, saving from the costly storage of the whole original matrix. A competition-based sparsification mechanism is further proposed to avoid the storage of tunable weight indexes. To maintain the effectiveness of sparse tuning with low-rank matrices, we extend the low-rank decomposition by applying nonlinear kernel functions to the whole-matrix merging. Consequently, we gain an increase in the rank of the merged matrix, enhancing the ability of SNELL in adapting the pre-trained models to downstream tasks. Extensive experiments on multiple downstream tasks show that SNELL achieves state-of-the-art performance with low memory usage, endowing PEFT with sparse tuning to large-scale models. Codes are available at https://github.com/ssfgunner/SNELL.

Towards Dynamic Message Passing on Graphs

Message passing plays a vital role in graph neural networks (GNNs) for effective feature learning. However, the over-reliance on input topology diminishes the efficacy of message passing and restricts the ability of GNNs. Despite efforts to mitigate the reliance, existing study encounters message-passing bottlenecks or high computational expense problems, which invokes the demands for flexible message passing with low complexity. In this paper, we propose a novel dynamic message-passing mechanism for GNNs. It projects graph nodes and learnable pseudo nodes into a common space with measurable spatial relations between them. With nodes moving in the space, their evolving relations facilitate flexible pathway construction for a dynamic message-passing process. Associating pseudo nodes to input graphs with their measured relations, graph nodes can communicate with each other intermediately through pseudo nodes under linear complexity. We further develop a GNN model named $\mathtt{\mathbf{N^2}}$ based on our dynamic message-passing mechanism. $\mathtt{\mathbf{N^2}}$ employs a single recurrent layer to recursively generate the displacements of nodes and construct optimal dynamic pathways. Evaluation on eighteen benchmarks demonstrates the superior performance of $\mathtt{\mathbf{N^2}}$ over popular GNNs. $\mathtt{\mathbf{N^2}}$ successfully scales to large-scale benchmarks and requires significantly fewer parameters for graph classification with the shared recurrent layer.

Prompting Video-Language Foundation Models with Domain-specific Fine-grained Heuristics for Video Question Answering

Video Question Answering (VideoQA) represents a crucial intersection between video understanding and language processing, requiring both discriminative unimodal comprehension and sophisticated cross-modal interaction for accurate inference. Despite advancements in multi-modal pre-trained models and video-language foundation models, these systems often struggle with domain-specific VideoQA due to their generalized pre-training objectives. Addressing this gap necessitates bridging the divide between broad cross-modal knowledge and the specific inference demands of VideoQA tasks. To this end, we introduce HeurVidQA, a framework that leverages domain-specific entity-action heuristics to refine pre-trained video-language foundation models. Our approach treats these models as implicit knowledge engines, employing domain-specific entity-action prompters to direct the model's focus toward precise cues that enhance reasoning. By delivering fine-grained heuristics, we improve the model's ability to identify and interpret key entities and actions, thereby enhancing its reasoning capabilities. Extensive evaluations across multiple VideoQA datasets demonstrate that our method significantly outperforms existing models, underscoring the importance of integrating domain-specific knowledge into video-language models for more accurate and context-aware VideoQA.

Multi-granularity Contrastive Cross-modal Collaborative Generation for End-to-End Long-term Video Question Answering

Long-term Video Question Answering (VideoQA) is a challenging vision-and-language bridging task focusing on semantic understanding of untrimmed long-term videos and diverse free-form questions, simultaneously emphasizing comprehensive cross-modal reasoning to yield precise answers. The canonical approaches often rely on off-the-shelf feature extractors to detour the expensive computation overhead, but often result in domain-independent modality-unrelated representations. Furthermore, the inherent gradient blocking between unimodal comprehension and cross-modal interaction hinders reliable answer generation. In contrast, recent emerging successful video-language pre-training models enable cost-effective end-to-end modeling but fall short in domain-specific ratiocination and exhibit disparities in task formulation. Toward this end, we present an entirely end-to-end solution for long-term VideoQA: Multi-granularity Contrastive cross-modal collaborative Generation (MCG) model. To derive discriminative representations possessing high visual concepts, we introduce Joint Unimodal Modeling (JUM) on a clip-bone architecture and leverage Multi-granularity Contrastive Learning (MCL) to harness the intrinsically or explicitly exhibited semantic correspondences. To alleviate the task formulation discrepancy problem, we propose a Cross-modal Collaborative Generation (CCG) module to reformulate VideoQA as a generative task instead of the conventional classification scheme, empowering the model with the capability for cross-modal high-semantic fusion and generation so as to rationalize and answer. Extensive experiments conducted on six publicly available VideoQA datasets underscore the superiority of our proposed method.

Suppress Content Shift: Better Diffusion Features via Off-the-Shelf Generation Techniques

Diffusion models are powerful generative models, and this capability can also be applied to discrimination. The inner activations of a pre-trained diffusion model can serve as features for discriminative tasks, namely, diffusion feature. We discover that diffusion feature has been hindered by a hidden yet universal phenomenon that we call content shift. To be specific, there are content differences between features and the input image, such as the exact shape of a certain object. We locate the cause of content shift as one inherent characteristic of diffusion models, which suggests the broad existence of this phenomenon in diffusion feature. Further empirical study also indicates that its negative impact is not negligible even when content shift is not visually perceivable. Hence, we propose to suppress content shift to enhance the overall quality of diffusion features. Specifically, content shift is related to the information drift during the process of recovering an image from the noisy input, pointing out the possibility of turning off-the-shelf generation techniques into tools for content shift suppression. We further propose a practical guideline named GATE to efficiently evaluate the potential benefit of a technique and provide an implementation of our methodology. Despite the simplicity, the proposed approach has achieved superior results on various tasks and datasets, validating its potential as a generic booster for diffusion features. Our code is available at https://github.com/Darkbblue/diffusion-content-shift.