Search

Although vision-language models (VLMs) have achieved significant success in various applications such as visual question answering, their resilience to prompt variations remains an under-explored area. Understanding how distractions affect VLMs is crucial for improving their real-world applicability, as inputs could have noisy and irrelevant information in many practical scenarios. This paper aims to assess the robustness of VLMs against both visual and textual distractions in the context of science question answering. Built on the ScienceQA dataset, we developed a new benchmark that introduces distractions in both the visual and textual contexts to evaluate the reasoning capacity of VLMs amid these distractions. Our findings reveal that most-of-the-art VLMs, including GPT-4, are vulnerable to various types of distractions, experiencing noticeable degradation in reasoning capabilities when confronted with distractions. Notably, models such as InternVL2 demonstrate a higher degree of robustness to these distractions. We also found that models exhibit greater sensitivity to textual distractions than visual ones. Additionally, we explored various mitigation strategies, such as prompt engineering, to counteract the impact of distractions. While these strategies improved solution accuracy, our analysis shows that there remain significant opportunities for improvement.

Unified multimodal large language models (U-MLLMs) have demonstrated impressive performance in visual understanding and generation in an end-to-end pipeline. Compared with generation-only models (e.g., Stable Diffusion), U-MLLMs may raise new questions about bias in their outputs, which can be affected by their unified capabilities. This gap is particularly concerning given the under-explored risk of propagating harmful stereotypes. In this paper, we benchmark the latest U-MLLMs and find that most exhibit significant demographic biases, such as gender and race bias. To better understand and mitigate this issue, we propose a locate-then-fix strategy, where we audit and show how the individual model component is affected by bias. Our analysis shows that bias originates primarily from the language model. More interestingly, we observe a "partial alignment" phenomenon in U-MLLMs, where understanding bias appears minimal, but generation bias remains substantial. Thus, we propose a novel balanced preference model to balance the demographic distribution with synthetic data. Experiments demonstrate that our approach reduces demographic bias while preserving semantic fidelity. We hope our findings underscore the need for more holistic interpretation and debiasing strategies of U-MLLMs in the future.

Different from the traditional semi-supervised learning paradigm that is constrained by the close-world assumption, Generalized Category Discovery (GCD) presumes that the unlabeled dataset contains new categories not appearing in the labeled set, and aims to not only classify old categories but also discover new categories in the unlabeled data. Existing studies on GCD typically devote to transferring the general knowledge from the self-supervised pretrained model to the target GCD task via some fine-tuning strategies, such as partial tuning and prompt learning. Nevertheless, these fine-tuning methods fail to make a sound balance between the generalization capacity of pretrained backbone and the adaptability to the GCD task. To fill this gap, in this paper, we propose a novel adapter-tuning-based method named AdaptGCD, which is the first work to introduce the adapter tuning into the GCD task and provides some key insights expected to enlighten future research. Furthermore, considering the discrepancy of supervision information between the old and new classes, a multi-expert adapter structure equipped with a route assignment constraint is elaborately devised, such that the data from old and new classes are separated into different expert groups. Extensive experiments are conducted on 7 widely-used datasets. The remarkable improvements in performance highlight the effectiveness of our proposals.

This paper explores concurrent FL processes within a three-tier system, with edge servers between edge devices and FL servers. A challenge in this setup is the limited bandwidth from edge devices to edge servers. Thus, allocating the bandwidth efficiently and fairly to support simultaneous FL processes becomes crucial. We propose a game-theoretic approach to model the bandwidth allocation problem and develop distributed and centralized heuristic schemes to find an approximate Nash Equilibrium of the game. We proposed the approach mentioned above using centralized and entirely distributed assumptions. Through rigorous analysis and experimentation, we demonstrate that our schemes efficiently and fairly assign the bandwidth to the FL processes for centralized and distributed solutions and outperform a baseline scheme where each edge server assigns bandwidth proportionally to the FL servers' requests that it receives. The proposed distributed and centralized schemes have comptetive performance.

Parameter-efficient tunings (PETs) have demonstrated impressive performance and promising perspectives in training large models, while they are still confronted with a common problem: the trade-off between learning new content and protecting old knowledge, leading to zero-shot generalization collapse, and cross-modal hallucination. In this paper, we reformulate Adapter, LoRA, Prefix-tuning, and Prompt-tuning from the perspective of gradient projection, and firstly propose a unified framework called Parameter Efficient Gradient Projection (PEGP). We introduce orthogonal gradient projection into different PET paradigms and theoretically demonstrate that the orthogonal condition for the gradient can effectively resist forgetting even for large-scale models. It therefore modifies the gradient towards the direction that has less impact on the old feature space, with less extra memory space and training time. We extensively evaluate our method with different backbones, including ViT and CLIP, on diverse datasets, and experiments comprehensively demonstrate its efficiency in reducing forgetting in class, online class, domain, task, and multi-modality continual settings. The project page is available at https://dmcv-ecnu-pegp.github.io/.

Large language models (LLMs) show excellent performance in difficult tasks, but they often require massive memories and computational resources. How to reduce the parameter scale of LLMs has become research hotspots. In this study, we make an important observation that the multi-head self-attention (MHA) sub-layer of Transformer exhibits noticeable low-rank structure, while the feed-forward network (FFN) sub-layer does not. With this regard, we design a mixed compression model, which organically combines Low-Rank matrix approximation And structured Pruning (LoRAP). For the MHA sub-layer, we propose an input activation weighted singular value decomposition method to strengthen the low-rank characteristic. Furthermore, we discover that the weight matrices in MHA sub-layer have different low-rank degrees. Thus, a novel parameter allocation scheme according to the discrepancy of low-rank degrees is devised. For the FFN sub-layer, we propose a gradient-free structured channel pruning method. During the pruning, we get an interesting finding that the least important 1% of parameter actually play a vital role in model performance. Extensive evaluations on zero-shot perplexity and zero-shot task classification indicate that our proposal is superior to previous structured compression rivals under multiple compression ratios., Comment: 8 pages,4 figures

Autoregressive decoding strategy is a commonly used method for text generation tasks with pre-trained language models, while early-exiting is an effective approach to speedup the inference stage. In this work, we propose a novel decoding strategy named Hierarchical Skip Decoding (HSD) for efficient autoregressive text generation. Different from existing methods that require additional trainable components, HSD is a plug-and-play method applicable to autoregressive text generation models, it adaptively skips decoding layers in a hierarchical manner based on the current sequence length, thereby reducing computational workload and allocating computation resources. Comprehensive experiments on five text generation datasets with pre-trained language models demonstrate HSD's advantages in balancing efficiency and text quality. With almost half of the layers skipped, HSD can sustain 90% of the text quality compared to vanilla autoregressive decoding, outperforming the competitive approaches.

This paper investigates the convergence rate for Tikhonov regularization of the problem of identifying the coefficient $a \in L^{\infty}(\Omega)$ in the Robin-boundary equation $-\mathrm{div}(a\nabla u)-bu=f,~ x \in \Omega \subset \mathbb R^M,~ M \geq 1$ and $u=0,~ x ~on~ \partial\Omega$, where $f(x)\in L^{\infty}(\Omega)$. Assume we only know the imprecise values of $u$ in the subset $\Omega_1 \subset \Omega$ given by $z^{\delta} \in {H}^1(\Omega_1)$, satisfies $\|u-z^{\delta}\|_{H^1(\Omega_1)}\leq \delta$. We assume $u$ satisfy the following boundary conditions on $\partial\Omega_1$: \begin{align*} \nabla u \cdot \vec{n}+\gamma u =0~on~\partial\Omega_1, \end{align*} where $\vec{n}$ is the normal vector of $\partial\Omega_1$ and $\gamma>0$ is a constant. We regularize this problem by correspondingly minimizing the strictly convex functional: \begin{align*} \min \limits_{a \in \mathbb A} &\frac12 \int_{\Omega_1} a | {\nabla(U(a)-z^\delta)}|^2 +\frac12\int_{\partial\Omega_1} a\gamma [U(a)-z^\delta]^2-\frac12 \int_{\Omega_1} b [U(a)-z^\delta]^2\\ &+ \rho \| a-a^* \|^2_{L^2(\Omega)}, \end{align*} where $U(a)$ is a map for $a$ to the solution of the Robin-boundary problem, $\rho > 0$ is the regularization parameter and $a^*$ is a priori estimate of $a$. We prove that the functional attain a unique global minimizer on the admissible set. Further, we give very simple source condition without the smallness requirement on the source function which provide the convergence rate $O(\sqrt{\delta})$ for the regularized solution.

With the popularity of cloud computing and machine learning, it has been a trend to outsource machine learning processes (including model training and model-based inference) to cloud. By the outsourcing, other than utilizing the extensive and scalable resource offered by the cloud service provider, it will also be attractive to users if the cloud servers can manage the machine learning processes autonomously on behalf of the users. Such a feature will be especially salient when the machine learning is expected to be a long-term continuous process and the users are not always available to participate. Due to security and privacy concerns, it is also desired that the autonomous learning preserves the confidentiality of users' data and models involved. Hence, in this paper, we aim to design a scheme that enables autonomous and confidential model refining in cloud. Homomorphic encryption and trusted execution environment technology can protect confidentiality for autonomous computation, but each of them has their limitations respectively and they are complementary to each other. Therefore, we further propose to integrate these two techniques in the design of the model refining scheme. Through implementation and experiments, we evaluate the feasibility of our proposed scheme. The results indicate that, with our proposed scheme the cloud server can autonomously refine an encrypted model with newly provided encrypted training data to continuously improve its accuracy. Though the efficiency is still significantly lower than the baseline scheme that refines plaintext-model with plaintext-data, we expect that it can be improved by fully utilizing the higher level of parallelism and the computational power of GPU at the cloud server., Comment: IEEE INTERNATIONAL CONFERENCE ON CLOUD COMPUTING (CLOUD) 2023

The increasing size of language models raises great research interests in parameter-efficient fine-tuning such as LoRA that freezes the pre-trained model, and injects small-scale trainable parameters for multiple downstream tasks (e.g., summarization, question answering and translation). To further enhance the efficiency of fine-tuning, we propose a framework that integrates LoRA and structured layer pruning. The integrated framework is validated on two created deidentified medical report summarization datasets based on MIMIC-IV-Note and two public medical dialogue datasets. By tuning 0.6% parameters of the original model and pruning over 30% Transformer-layers, our framework can reduce 50% of GPU memory usage and speed up 100% of the training phase, while preserving over 92% generation qualities on free-text sequence-to-sequence tasks.

Self-supervised skeleton-based action recognition enjoys a rapid growth along with the development of contrastive learning. The existing methods rely on imposing invariance to augmentations of 3D skeleton within a single data stream, which merely leverages the easy positive pairs and limits the ability to explore the complicated movement patterns. In this paper, we advocate that the defect of single-stream contrast and the lack of necessary feature transformation are responsible for easy positives, and therefore propose a Cross-Stream Contrastive Learning framework for skeleton-based action Representation learning (CSCLR). Specifically, the proposed CSCLR not only utilizes intra-stream contrast pairs, but introduces inter-stream contrast pairs as hard samples to formulate a better representation learning. Besides, to further exploit the potential of positive pairs and increase the robustness of self-supervised representation learning, we propose a Positive Feature Transformation (PFT) strategy which adopts feature-level manipulation to increase the variance of positive pairs. To validate the effectiveness of our method, we conduct extensive experiments on three benchmark datasets NTU-RGB+D 60, NTU-RGB+D 120 and PKU-MMD. Experimental results show that our proposed CSCLR exceeds the state-of-the-art methods on a diverse range of evaluation protocols., Comment: 15 pages, 7 figures

Domain adaptation manages to transfer the knowledge of well-labeled source data to unlabeled target data. Many recent efforts focus on improving the prediction accuracy of target pseudo-labels to reduce conditional distribution shift. In this paper, we propose a novel domain adaptation method, which infers target pseudo-labels through cross-domain label propagation, such that the underlying manifold structure of two domain data can be explored. Unlike existing cross-domain label propagation methods that separate domain-invariant feature learning, affinity matrix constructing and target labels inferring into three independent stages, we propose to integrate them into a unified optimization framework. In such way, these three parts can boost each other from an iterative optimization perspective and thus more effective knowledge transfer can be achieved. Furthermore, to construct a high-quality affinity matrix, we propose a discriminative graph self-learning strategy, which can not only adaptively capture the inherent similarity of the data from two domains but also effectively exploit the discriminative information contained in well-labeled source data and pseudo-labeled target data. An efficient iterative optimization algorithm is designed to solve the objective function of our proposal. Notably, the proposed method can be extended to semi-supervised domain adaptation in a simple but effective way and the corresponding optimization problem can be solved with the identical algorithm. Extensive experiments on six standard datasets verify the significant superiority of our proposal in both unsupervised and semi-supervised domain adaptation settings.

This study presents three deidentified large medical text datasets, named DISCHARGE, ECHO and RADIOLOGY, which contain 50K, 16K and 378K pairs of report and summary that are derived from MIMIC-III, respectively. We implement convincing baselines of automated abstractive summarization on the proposed datasets with pre-trained encoder-decoder language models, including BERT2BERT, T5-large and BART. Further, based on the BART model, we leverage the sampled summaries from the train set as prior knowledge guidance, for encoding additional contextual representations of the guidance with the encoder and enhancing the decoding representations in the decoder. The experimental results confirm the improvement of ROUGE scores and BERTScore made by the proposed method, outperforming the larger model T5-large.

As a recent noticeable topic, domain generalization aims to learn a generalizable model on multiple source domains, which is expected to perform well on unseen test domains. Great efforts have been made to learn domain-invariant features by aligning distributions across domains. However, existing works are often designed based on some relaxed conditions which are generally hard to satisfy and fail to realize the desired joint distribution alignment. In this paper, we propose a novel domain generalization method, which originates from an intuitive idea that a domain-invariant classifier can be learned by minimizing the KL-divergence between posterior distributions from different domains. To enhance the generalizability of the learned classifier, we formalize the optimization objective as an expectation computed on the ground-truth marginal distribution. Nevertheless, it also presents two obvious deficiencies, one of which is the side-effect of entropy increase in KL-divergence and the other is the unavailability of ground-truth marginal distributions. For the former, we introduce a term named maximum in-domain likelihood to maintain the discrimination of the learned domain-invariant representation space. For the latter, we approximate the ground-truth marginal distribution with source domains under a reasonable convex hull assumption. Finally, a Constrained Maximum Cross-domain Likelihood (CMCL) optimization problem is deduced, by solving which the joint distributions are naturally aligned. An alternating optimization strategy is carefully designed to approximately solve this optimization problem. Extensive experiments on four standard benchmark datasets, i.e., Digits-DG, PACS, Office-Home and miniDomainNet, highlight the superior performance of our method.

Domain generalization (DG) aims to learn a model on several source domains, hoping that the model can generalize well to unseen target domains. The distribution shift between domains contains the covariate shift and conditional shift, both of which the model must be able to handle for better generalizability. In this paper, a novel DG method is proposed to deal with the distribution shift via Visual Alignment and Uncertainty-guided belief Ensemble (VAUE). Specifically, for the covariate shift, a visual alignment module is designed to align the distribution of image style to a common empirical Gaussian distribution so that the covariate shift can be eliminated in the visual space. For the conditional shift, we adopt an uncertainty-guided belief ensemble strategy based on the subjective logic and Dempster-Shafer theory. The conditional distribution given a test sample is estimated by the dynamic combination of that of source domains. Comprehensive experiments are conducted to demonstrate the superior performance of the proposed method on four widely used datasets, i.e., Office-Home, VLCS, TerraIncognita, and PACS.

Temporal Action Localization (TAL) aims to predict both action category and temporal boundary of action instances in untrimmed videos, i.e., start and end time. Fully-supervised solutions are usually adopted in most existing works, and proven to be effective. One of the practical bottlenecks in these solutions is the large amount of labeled training data required. To reduce expensive human label cost, this paper focuses on a rarely investigated yet practical task named semi-supervised TAL and proposes an effective active learning method, named AL-STAL. We leverage four steps for actively selecting video samples with high informativeness and training the localization model, named \emph{Train, Query, Annotate, Append}. Two scoring functions that consider the uncertainty of localization model are equipped in AL-STAL, thus facilitating the video sample rank and selection. One takes entropy of predicted label distribution as measure of uncertainty, named Temporal Proposal Entropy (TPE). And the other introduces a new metric based on mutual information between adjacent action proposals and evaluates the informativeness of video samples, named Temporal Context Inconsistency (TCI). To validate the effectiveness of proposed method, we conduct extensive experiments on two benchmark datasets THUMOS'14 and ActivityNet 1.3. Experiment results show that AL-STAL outperforms the existing competitors and achieves satisfying performance compared with fully-supervised learning., Comment: Need to modify

In group activity recognition, hierarchical framework is widely adopted to represent the relationships between individuals and their corresponding group, and has achieved promising performance. However, the existing methods simply employed max/average pooling in this framework, which ignored the distinct contributions of different individuals to the group activity recognition. In this paper, we propose a new contextual pooling scheme, named attentive pooling, which enables the weighted information transition from individual actions to group activity. By utilizing the attention mechanism, the attentive pooling is intrinsically interpretable and able to embed member context into the existing hierarchical model. In order to verify the effectiveness of the proposed scheme, two specific attentive pooling methods, i.e., global attentive pooling (GAP) and hierarchical attentive pooling (HAP) are designed. GAP rewards the individuals that are significant to group activity, while HAP further considers the hierarchical division by introducing subgroup structure. The experimental results on the benchmark dataset demonstrate that our proposal is significantly superior beyond the baseline and is comparable to the state-of-the-art methods., Comment: 7 pages, 7 figures

Information Bottleneck (IB) based multi-view learning provides an information theoretic principle for seeking shared information contained in heterogeneous data descriptions. However, its great success is generally attributed to estimate the multivariate mutual information which is intractable when the network becomes complicated. Moreover, the representation learning tradeoff, {\it i.e.}, prediction-compression and sufficiency-consistency tradeoff, makes the IB hard to satisfy both requirements simultaneously. In this paper, we design several variational information bottlenecks to exploit two key characteristics ({\it i.e.}, sufficiency and consistency) for multi-view representation learning. Specifically, we propose a Multi-View Variational Distillation (MV$^2$D) strategy to provide a scalable, flexible and analytical solution to fitting MI by giving arbitrary input of viewpoints but without explicitly estimating it. Under rigorously theoretical guarantee, our approach enables IB to grasp the intrinsic correlation between observations and semantic labels, producing predictive and compact representations naturally. Also, our information-theoretic constraint can effectively neutralize the sensitivity to heterogeneous data by eliminating both task-irrelevant and view-specific information, preventing both tradeoffs in multiple view cases. To verify our theoretically grounded strategies, we apply our approaches to various benchmarks under three different applications. Extensive experiments to quantitatively and qualitatively demonstrate the effectiveness of our approach against state-of-the-art methods.

Multi-view clustering has attracted much attention thanks to the capacity of multi-source information integration. Although numerous advanced methods have been proposed in past decades, most of them generally overlook the significance of weakly-supervised information and fail to preserve the feature properties of multiple views, thus resulting in unsatisfactory clustering performance. To address these issues, in this paper, we propose a novel Deep Multi-view Semi-supervised Clustering (DMSC) method, which jointly optimizes three kinds of losses during networks finetuning, including multi-view clustering loss, semi-supervised pairwise constraint loss and multiple autoencoders reconstruction loss. Specifically, a KL divergence based multi-view clustering loss is imposed on the common representation of multi-view data to perform heterogeneous feature optimization, multi-view weighting and clustering prediction simultaneously. Then, we innovatively propose to integrate pairwise constraints into the process of multi-view clustering by enforcing the learned multi-view representation of must-link samples (cannot-link samples) to be similar (dissimilar), such that the formed clustering architecture can be more credible. Moreover, unlike existing rivals that only preserve the encoders for each heterogeneous branch during networks finetuning, we further propose to tune the intact autoencoders frame that contains both encoders and decoders. In this way, the issue of serious corruption of view-specific and view-shared feature space could be alleviated, making the whole training procedure more stable. Through comprehensive experiments on eight popular image datasets, we demonstrate that our proposed approach performs better than the state-of-the-art multi-view and single-view competitors.

When neural network model and data are outsourced to cloud server for inference, it is desired to preserve the confidentiality of model and data as the involved parties (i.e., cloud server, model providing client and data providing client) may not trust mutually. Solutions were proposed based on multi-party computation, trusted execution environment (TEE) and leveled or fully homomorphic encryption (LHE/FHE), but their limitations hamper practical application. We propose a new framework based on synergistic integration of LHE and TEE, which enables collaboration among mutually-untrusted three parties, while minimizing the involvement of (relatively) resource-constrained TEE and allowing the full utilization of the untrusted but more resource-rich part of server. We also propose a generic and efficient LHE-based inference scheme as an important performance-determining component of the framework. We implemented/evaluated the proposed system on a moderate platform and show that, our proposed scheme is more applicable/scalable to various settings, and has better performance, compared to the state-of-the-art LHE-based solutions.