United States

Current multimodal sentiment analysis (MSA) and emotion recognition in conversations (ERC) methods based on pre-trained language models exhibit two primary limitations:
1) Once trained for MSA and ERC tasks, these pre-trained language models lose their original generalized capabilities. 2) They demand considerable computational resources.  As the size of pre-trained language models continues to grow, training larger multimodal sentiment analysis models using previous approaches could result in unnecessary computational cost.  In response to this challenge, we propose Multimodal Sentiment Analysis and Emotion Recognition Adapter (MSE-Adapter), a lightweight and adaptable plugin. This plugin enables a large language model (LLM) to carry out MSA or ERC tasks with minimal computational overhead (only introduces approximately 2.6M to 2.8M trainable parameters upon the 6/7B models),  while preserving the intrinsic capabilities of the LLM. In the MSE-Adapter, the Text-Guide-Mixer (TGM) module is introduced to establish explicit connections between non-textual and textual modalities through the Hadamard product. This allows non-textual modalities to better align with textual modalities at the feature level, promoting the generation of higher-quality pseudo tokens. Extensive experiments were conducted on four public English and Chinese datasets using consumer-grade GPUs and open-source LLMs (Qwen-1.8B, ChatGLM3-6B-base, and LLaMA2-7B) as the backbone. The results demonstrate the effectiveness of the proposed plugin. The code will be released on GitHub after a blind review.

AAAI 2025

MSE-Adapter: A Lightweight Plugin Endowing LLMs with the Capability to Perform Multimodal Sentiment Analysis and Emotion Recognition

snlp

multi-modal nlp

language grounding

poster

We are pleased to announce the Thirty-Ninth AAAI Conference on Artificial Intelligence (AAAI-25), which will be held in Philadelphia, Pennsylvania at the Pennsylvania Convention Center from February 25 to March 4, 2025.

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-25 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and a range of other activities to be announced.

### [Invited Speakers](https://aaai.org/conference/aaai/aaai-25/aaai-25-invited-speakers/)

Register [here](https://aaai.org/conference/aaai/aaai-25/registration/)

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-25 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and a range of other activities to be announced.



The strength of multimodal learning lies in its ability to integrate information from various sources, providing rich and comprehensive insights. However, in real-world scenarios, multi-modal systems often face the challenge of dynamic modality contributions, the dominance of different modalities may change with the environments, leading to suboptimal performance in multimodal learning. Current methods mainly enhance weak modalities to balance multimodal representation bias, which inevitably optimizes from a partial modality perspective, easily leading to performance descending for dominant modalities. To address this problem, we propose an Asymmetric Reinforcing method against Multimodal Representation bias (ARMR). Our ARMR dynamically reinforces on the weak modalities while maintaining the representation ability of dominant modalities through condition mutual information. Moreover, we first provide an in-depth analysis that optimizing certain modalities could cause information loss and prevent leveraging the full advantages of multimodal data. By exploring the dominance and narrowing the contribution gaps between modalities, we have significantly improved the performance of multimodal learning, making notable progress in mitigating imbalanced multimodal cooperation.

Asymmetric Reinforcing against Multi-modal Representation Bias

With the explosive growth of deep learning applications, the right to be forgotten has been required unprecedentedly in various AI industries. For example, given a facial recognition system, some individuals may wish to remove images that might have been used in the training phase from the trained model. Unfortunately, modern deep neural networks sometimes unexpectedly leak personal identities.
Recent studies have presented various machine unlearning algorithms to make a trained model unlearn the data to be forgotten. 
While these methods generally perform well in terms of forgetting scores, we have found that an unexpected model utility drop can occur. This phenomenon, which we term correlation collapse, happens when the machine unlearning algorithms reduce the useful correlation between image features and the true label. To address this challenge, we propose Distribution-Level Feature Distancing (DLFD), a novel method that efficiently forgets instances while preventing correlation collapse. Our method synthesizes data samples so that the generated data distribution is far from the distribution of samples being forgotten in the feature space, achieving significant results even within a single training epoch. Through extensive experiments on facial recognition datasets, we demonstrate that our approach significantly outperforms state-of-the-art machine unlearning methods. All datasets and codes will be publicly available upon acceptance of the paper. All datasets and codes are publicly available after they are accepted.

Distribution-Level Feature Distancing for Machine Unlearning: Towards a Better Trade-off Between Model Utility and Forgetting

Blind-spot networks (BSN) have been prevalent neural architectures in self-supervised image denoising (SSID). However, most existing BSNs are conducted with convolution layers. Although transformers have shown the potential to overcome the limitations of convolutions in many image restoration tasks, the attention mechanisms may violate the blind-spot requirement, thereby restricting their applicability in BSN. To this end, we propose to analyze and redesign the channel and spatial attentions to meet the blind-spot requirement. Specifically, channel self-attention may leak the blind-spot information in multi-scale architectures, since the downsampling shuffles the spatial feature into channel dimensions. To alleviate this problem, we divide the channel into several groups and perform channel attention separately. For spatial self-attention, we apply an elaborate mask to the attention matrix to restrict and mimic the receptive field of dilated convolution. Based on the redesigned channel and window attentions, we build a Transformer-based Blind-Spot Network (TBSN), which shows strong local fitting and global perspective abilities. Furthermore, we introduce a knowledge distillation strategy that distills TBSN into smaller denoisers to improve computational efficiency while maintaining performance. Extensive experiments on real-world image denoising datasets show that TBSN largely extends the receptive field and exhibits favorable performance against state-of-the-art SSID methods. The code and pre-trained models will be publicly available.

Rethinking Transformer-Based Blind-Spot Network for Self-Supervised Image Denoising

Model extraction attacks are one type of inference-time attacks that approximate the functionality and performance of a black-box victim model by launching a certain number of queries to the model and then leveraging the model's predictions to train a substitute model. 
These attacks pose severe security threats to production models and MLaaS platforms and could cause significant monetary losses to the model owners. 
A body of work has proposed to defend machine learning models against model extraction attacks, including both active defense methods that modify the model's outputs or increase the query overhead to avoid extraction and passive defense methods that detect malicious queries or leverage watermarks to perform post-verification.
In this work, we introduce a new defense paradigm called attack as defense which modifies the model's output to be poisonous such that any malicious users that attempt to use the output to train a substitute model will be poisoned. 
To this end, we propose a novel lightweight backdoor attack method dubbed HoneypotNet that replaces the classification layer of the victim model with a honeypot layer and then finetunes the honeypot layer with a shadow model (to simulate model extraction) via bi-level optimization to modify its output to be poisonous while remaining the original performance. 
We empirically demonstrate on four commonly used benchmark datasets that HoneypotNet can inject backdoors into substitute models with a high success rate. 
The injected backdoor not only facilitates ownership verification but also disrupts the functionality of substitute models, serving as a significant deterrent to model extraction attacks.

HoneypotNet: Backdoor Attacks Against Model Extraction

Human beings often experience stress, which can significantly influence their performance. This study explores whether Large Language Models (LLMs) exhibit stress responses similar to those of humans and whether their performance fluctuates under different stress-inducing prompts. To investigate this, we developed a novel set of prompts, termed \textit{StressPrompt}, designed to induce varying levels of stress. These prompts were derived from established psychological frameworks and carefully calibrated based on ratings from human participants. We then applied these prompts to several LLMs to assess their responses across a range of tasks, including instruction-following, complex reasoning, and emotional intelligence. The findings suggest that LLMs, like humans, perform optimally under moderate stress, consistent with the Yerkes-Dodson law. Notably, their performance declines under both low and high-stress conditions. Our analysis further revealed that these \textit{StressPrompts} significantly alter the internal states of LLMs, leading to changes in their neural representations that mirror human responses to stress. This research provides critical insights into the operational robustness and flexibility of LLMs, demonstrating the importance of designing AI systems capable of maintaining high performance in real-world scenarios where stress is prevalent, such as in customer service, healthcare, and emergency response contexts. Moreover, this study contributes to the broader AI research community by offering a new perspective on how LLMs handle different scenarios and their similarities to human cognition.

StressPrompt: Does Stress Impact Large Language Models and Human Performance Similarly?

Transformers have revolutionized the point cloud learning task, but the quadratic complexity hinders its extension to long sequence and makes a burden on limited computational resources. The recent advent of RWKV, a fresh breed of deep sequence models, has shown immense potential for sequence modeling in NLP tasks. In this paper, we present PointRWKV, a model of linear complexity derived from the RWKV model in the NLP field with necessary modifications for point cloud learning tasks. Specifically, taking the embedded point patches as input, we first propose to explore the global processing capabilities within PointRWKV blocks using modified multi-headed matrix-valued states and a dynamic attention recurrence mechanism. To extract local geometric features simultaneously, we design a  parallel branch to encode the point cloud efficiently in a fixed radius near-neighbors graph with a graph stabilizer. Furthermore, we design PointRWKV as a multi-scale framework for hierarchical feature learning of 3D point clouds, facilitating various downstream tasks. Extensive experiments on different point cloud learning tasks show our proposed PointRWKV outperforms the transformer- and mamba-based counterparts, while significantly saving about 42\% FLOPs, demonstrating the potential option for constructing foundational 3D models. 
The code and models will be available for further research.

PointRWKV: Efficient RWKV-Like Model for Hierarchical Point Cloud Learning

Chart Question Answering (CQA) requires models to perform chart perception and reasoning. Recent studies driven by Large Language Models (LLMs) have dominated CQA. These include employing more cognitively capable LLMs for indirectly reasoning over transformed charts, i.e., tables, and directly perceiving charts utilizing Multimodal Large Language Models (MLLMs) with a wider perceptual range. Yet, they often encounter bottlenecks due to the limitation of the receptive field of LLMs and the fragility of the complex reasoning of MLLMs. To unite the strengths of LLMs and MLLMs to complement each other's limitations, we propose Synergy, a framework that unites the power of both models for CQA. Synergy first unites the chart with a table as the augmented perceptual signal. Next, it unites LLMs and MLLMs, scheduling the former to decompose a question into subquestions and the latter to answer these by perceiving the chart. Lastly, it operates LLMs to summarize the subquestion-answer pairs to refine the final answer. Extensive experimental results on popular CharQA and PlotQA benchmarks reveal that, with the power of union, Synergy outperforms strong competitors and achieves superior boosts over naive MLLMs by uniting them with a smaller LLM.

Union is Strength! Unite the Power of LLMs and MLLMs for Chart Question Answering

Person re-identification (re-ID) via 3D skeleton data is a challenging task with significant value in many scenarios. Existing skeleton-based methods typically assume virtual motion relations between all joints, and adopt average joint or sequence representations for learning. However, they rarely explore key body structure and motion such as gait to focus on more important body joints or limbs, while lacking the ability to fully mine valuable spatial-temporal sub-patterns of skeletons to enhance model learning. This paper presents a generic Motif guided graph transformer with Combinatorial skeleton prototype learning (MoCos) that exploits structure-specific and gait-related body relations as well as combinatorial features of skeleton graphs to learn effective skeleton representations for person re-ID. In particular, motivated by the locality within joints' structure and the body-component collaboration in gait, we first propose the motif guided graph transformer (MGT) that incorporates hierarchical structural motifs and gait collaborative motifs, which simultaneously focuses on multi-order local joint correlations and key cooperative body parts to enhance skeleton relation learning. Then, we devise the combinatorial skeleton prototype learning (CSP) that leverages random spatial-temporal combinations of joint nodes and skeleton graphs to generate diverse sub-skeleton and sub-tracklet representations, which are contrasted with the most representative features (prototypes) of each identity to learn class-related semantics and discriminative skeleton representations. Extensive experiments validate the superior performance of MoCos over existing state-of-the-art models. We further show its generality under RGB-estimated skeletons, different graph modeling, and unsupervised scenarios.

Motif Guided Graph Transformers with Combinatorial Skeleton Prototype Learning for Skeleton-Based Person Re-Identification

Large Language Model-based Dense Retrieval (LLM-DR) optimizes over numerous heterogeneous fine-tuning collections from different domains. However, the discussion about its training data distribution is still minimal. Previous studies rely on empirically assigned dataset choices or sampling ratios, which inevitably leads to sub-optimal retrieval performances. In this paper, we propose a new task-level Distributionally Robust Optimization (tDRO) algorithm for LLM-DR fine-tuning, targeted at improving the universal domain generalization ability by end-to-end reweighting the data distribution of each task. The tDRO parameterizes the domain weights and updates them with scaled domain gradients. The optimized weights are then transferred to the LLM-DR fine-tuning to train more robust retrievers. Experiments show optimal improvements in large-scale retrieval benchmarks and reduce up to 30\% dataset usage after applying our optimization algorithm with a series of different-sized LLM-DR models.

Task-level Distributionally Robust Optimization for Large Language Model-based Dense Retrieval

Visual storytelling involves generating a sequence of coherent frames from a textual storyline while maintaining consistency in characters and scenes. Existing autoregressive methods, which rely on previous frame-sentence pairs, struggle with high memory usage, slow generation speeds, and limited context integration. To address these issues, we propose ContextualStory, a novel framework designed to generate coherent story frames and extend frames for story continuation. ContextualStory utilizes Spatially-Enhanced Temporal Attention to capture spatial and temporal dependencies, handling significant character movements effectively. Additionally, we introduces a Storyline Contextualizer to enrich context in storyline embedding and a StoryFlow Adapter to measure scene changes between frames for guiding model. Extensive experiments on PororoSV and FlintstonesSV benchmarks demonstrate that ContextualStory significantly outperforms existing methods in both story visualization and story continuation.

Premium content

Next from AAAI 2025

Asymmetric Reinforcing against Multi-modal Representation Bias

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES