IJCAI 2024 Papers

Large Language Models (LLMs) have seen significant use in domains such as natural language processing and computer vision. Going beyond text, image and graphics, LLMs present a significant potential for analysis of time series data, benefiting domains such as climate, IoT, healthcare, traffic, audio and finance. This survey paper provides an in-depth exploration and a detailed taxonomy of the various methodologies employed to harness the power of LLMs for time series analysis. We address the inherent challenge of bridging the gap between LLMs' original text data training and the numerical nature of time series data, and explore strategies for transferring and distilling knowledge from LLMs to numerical time series analysis. We detail various methodologies, including (1) direct prompting of LLMs, (2) time series quantization, (3) aligning techniques, (4) utilization of the vision modality as a bridging mechanism, and (5) the combination of LLMs with tools. Additionally, this survey offers a comprehensive overview of the existing multimodal time series and text datasets in diverse domains, and discusses the challenges and future opportunities of this emerging field.

This paper focuses on the potential of Generative Artificial Intelligence (AI), particularly Large Language Models (LLMs), in the still unexplored domain of robotic dance creation. In particular, we assess whether a LLM (GPT-3.5 turbo) can create robotic dance choreographies, and we investigate if the feedback provided by human creators can improve the quality of the output. To this end, we design three prompt engineering techniques for robotic dance creation. In the prompts, we gradually introduce human knowledge through examples and feedback in natural language in order to explore the dynamics of human-AI co-creation. The experimental analysis shows that the capabilities of the LLM can be improved through human collaboration, by producing choreographies with a major artistic impact on the evaluation audience. The findings offer valuable insights into the interplay between human creativity and AI generative models, paving the way for enhanced collaborative frameworks in creative domains.

Large Language Models have shown tremendous performance on a large variety of natural language processing tasks, ranging from text comprehension to common sense reasoning. However, the mechanisms responsible for this success remain opaque, and it is unclear whether LLMs can achieve human-like cognitive capabilities or whether these models are still fundamentally circumscribed. Abstract reasoning is a fundamental task for cognition, consisting of finding and applying a general pattern from few data. Evaluating deep neural architectures on this task could give insight into their potential limitations regarding reasoning and their broad generalisation abilities, yet this is currently an under-explored area. In this paper, we introduce a new benchmark for evaluating language models beyond memorization on abstract reasoning tasks. We perform extensive evaluations of state-of-the-art LLMs, showing that they currently achieve very limited performance in contrast with other natural language tasks, even when applying techniques that have been shown to improve performance on other NLP tasks. We argue that guiding LLM generation to follow causal paths could help improve the generalisation and reasoning abilities of LLMs.

Algorithm selection, a critical process of automated machine learning, aims to identify the most suitable algorithm for solving a specific problem prior to execution. Mainstream algorithm selection techniques heavily rely on problem features, while the role of algorithm features remains largely unexplored. Due to the intrinsic complexity of algorithms, effective methods for universally extracting algorithm information are lacking. This paper takes a significant step towards bridging this gap by introducing Large Language Models (LLMs) into algorithm selection for the first time. By comprehending the code text, LLM not only captures the structural and semantic aspects of the algorithm, but also demonstrates contextual awareness and library function understanding. The high-dimensional algorithm representation extracted by LLM, after undergoing a feature selection module, is combined with the problem representation and passed to the similarity calculation module. The selected algorithm is determined by the matching degree between a given problem and different algorithms. Extensive experiments validate the performance superiority of the proposed model and the efficacy of each key module. Furthermore, we present a theoretical upper bound on model complexity, showcasing the influence of algorithm representation and feature selection modules. This provides valuable theoretical guidance for the practical implementation of our method.

Recent studies have uncovered the potential of Large Language Models (LLMs) in addressing complex sequential decision-making tasks through the provision of high-level instructions. However, LLM-based agents lack specialization in tackling specific target problems, particularly in real-time dynamic environments. Additionally, deploying an LLM-based agent in practical scenarios can be both costly and time-consuming. On the other hand, reinforcement learning (RL) approaches train agents that specialize in the target task but often suffer from low sampling efficiency and high exploration costs. In this paper, we introduce a novel framework that addresses these challenges by training a smaller, specialized student RL agent using instructions from an LLM-based teacher agent. By incorporating the guidance from the teacher agent, the student agent can distill the prior knowledge of the LLM into its own model. Consequently, the student agent can be trained with significantly less data. Moreover, through further training with environment feedback, the student agent surpasses the capabilities of its teacher for completing the target task. We conducted experiments on challenging MiniGrid and Habitat environments, specifically designed for embodied AI research, to evaluate the effectiveness of our framework. The results clearly demonstrate that our approach achieves superior performance compared to strong baseline methods. Our code is available at https://github.com/ZJLAB-AMMI/LLM4Teach.

Self-supervised methods have gained prominence in time series anomaly detection due to the scarcity of available annotations. Nevertheless, they typically demand extensive training data to acquire a generalizable representation map, which conflicts with scenarios of a few available samples, thereby limiting their performance. To overcome the limitation, we propose AnomalyLLM, a knowledge distillation-based time series anomaly detection approach where the student network is trained to mimic the features of the large language model (LLM)-based teacher network that is pretrained on large-scale datasets. During the testing phase, anomalies are detected when the discrepancy between the features of the teacher and student networks is large. To circumvent the student network from learning the teacher network’s feature of anomalous samples, we devise two key strategies. 1) Prototypical signals are incorporated into the student network to consolidate the normal feature extraction. 2) We use synthetic anomalies to enlarge the representation gap between the two networks. AnomalyLLM demonstrates state-of-the-art performance on 15 datasets, improving accuracy by at least 14.5% in the UCR dataset.

Large Language Models (LLMs) have achieved remarkable success across a wide array of tasks. Due to their notable capabilities in planning and reasoning, LLMs have been utilized as autonomous agents for the automatic execution of various tasks. Recently, LLM-based agent systems have rapidly evolved from single-agent planning or decision-making to operating as multi-agent systems, enhancing their ability in complex problem-solving and world simulation. To offer an overview of this dynamic field, we present this survey to offer an in-depth discussion on the essential aspects and challenges of LLM-based multi-agent (LLM-MA) systems. Our objective is to provide readers with an in-depth understanding of these key points: the domains and settings where LLM-MA systems operate or simulate; the profiling and communication methods of these agents; and the means by which these agents develop their skills. For those interested in delving into this field, we also summarize the commonly used datasets or benchmarks. To keep researchers updated on the latest studies, we maintain an open-source GitHub repository (github.com/taichengguo/LLM_MultiAgents_Survey_Papers), dedicated to outlining the research of LLM-MA research.

Current languages for specifying multiagent protocols either over-constrain protocol enactments or complicate capturing their meanings. We propose Langshaw, a declarative protocol language based on (1) sayso, a new construct that captures who has priority over setting each attribute, and (2) nono and nogo, two constructs to capture conflicts between actions. Langshaw combines flexibility with an information model to express meaning. We give a formal semantics for Langshaw, procedures for determining the safety and liveness of a protocol, and a method to generate a message-oriented protocol (embedding needed coordination) suitable for flexible asynchronous enactment.

LangXAI is a framework that integrates Explainable Artificial Intelligence (XAI) with advanced vision models to generate textual explanations for visual recognition tasks. Despite XAI advancements, an understanding gap persists for end-users with limited domain knowledge in artificial intelligence and computer vision. LangXAI addresses this by furnishing text-based explanations for classification, object detection, and semantic segmentation model outputs to end-users. Preliminary results demonstrate LangXAI's enhanced plausibility, with high BERTScore across tasks, fostering a more transparent and reliable AI framework on vision tasks for end-users. The code and demo of this work can be found at https://analytics-everywhere-lab.github.io/langxai.io/.

Semantic scene completion aims to infer the 3D geometric structures with semantic classes from camera or LiDAR, which provide essential occupancy information in autonomous driving. Prior endeavors concentrate on constructing the network or benchmark in a fully supervised manner. While the dense occupancy grids need point-wise semantic annotations, which incur expensive and tedious labeling costs. In this paper, we build a new label-efficient benchmark, named ScribbleSC, where the sparse scribble-based semantic labels are combined with dense geometric labels for semantic scene completion. In particular, we propose a simple yet effective approach called Scribble2Scene, which bridges the gap between the sparse scribble annotations and fully-supervision. Our method consists of geometric-aware auto-labelers construction and online model training with an offline-to-online distillation module to enhance the performance. Experiments on SemanticKITTI demonstrate that Scribble2Scene achieves competitive performance against the fully-supervised counterparts, showing 99% performance of the fully-supervised models with only 13.5% voxels labeled. Both annotations of ScribbleSC and our full implementation are available at https://github.com/songw-zju/Scribble2Scene.

Vertical federated learning (VFL) is a distributed machine learning paradigm that collaboratively trains models using passive parties with features and an active party with additional labels. While VFL offers privacy preservation through data localization, the threat of label leakage remains a significant challenge. Label leakage occurs due to label inference attacks, where passive parties attempt to infer labels for their privacy and commercial value. Extensive research has been conducted on this specific VFL attack, but a comprehensive summary is still lacking. To bridge this gap, our paper aims to survey the existing label inference attacks and defenses. We propose two new taxonomies for both label inference attacks and defenses, respectively. Beyond summarizing the current state of research, we highlight techniques that we believe hold potential and could significantly influence future studies. Moreover, experimental benchmark datasets and evaluation metrics are summarized to provide a guideline for subsequent work.

Label distribution learning (LDL) is an effective method to predict the label description degree (a.k.a. label distribution) of a sample. However, annotating label distribution (LD) for training samples is extremely costly. So recent studies often first use label enhancement (LE) to generate the estimated label distribution from the logical label and then apply external LDL algorithms on the recovered label distribution to predict the label distribution for unseen samples. But this step-wise manner overlooks the possible connections between LE and LDL. Moreover, the existing LE approaches may assign some description degrees to invalid labels. To solve the above problems, we propose a novel method to learn an LDL model directly from the logical label, which unifies LE and LDL into a joint model, and avoids the drawbacks of the previous LE methods. We also give the generalization error bound of our method and theoretically prove that directly learning an LDL model from the logical labels is feasible. Extensive experiments on various datasets prove that the proposed approach can construct a reliable LDL model directly from the logical label, and produce more accurate label distribution than the state-of-the-art LE methods. The code and the supplementary file can be found in https://github.com/seutjw/DLDL.

Graph-based consistency principle has been successfully applied to many semi-supervised problems in machine learning. Its performance largely depends on the quality of augmented graphs, which has been recently proven that revealing graph properties and maintaining the invariance of graphs are crucial for good performance. However, existing topology- or feature-based augmentation methods are spectrally non-localized -- important spectrums are disturbed throughout the entire frequency range, and their invariance may not be well preserved. Efforts on this issue remain to be limited. This paper proposes a simple yet effective model called Localized SPectral AugmentatioN (LSPAN), which perturbs a concentrated part of graph spectrum with equivalent intensity using Fourier orthogonality, so as to enhance graph spectrum preservation as well as model prediction. Moreover, it also avoids the significant training time of inverse Fourier transform. Extensive empirical evaluation on real-world datasets clearly shows the performance gain of spectrally localized augmentation, as well as its good convergence and efficiency compared to existing graph methods.

Self-correction is emerging as a promising approach to mitigate the issue of hallucination in Large Language Models (LLMs). To facilitate effective self-correction, recent research has proposed mistake detection as its initial step. However, current literature suggests that LLMs often struggle with reliably identifying reasoning mistakes when using simplistic prompting strategies. To address this challenge, we introduce a unique prompting strategy, termed the Pedagogical Chain-of-Thought (PedCoT), which is specifically designed to guide the identification of reasoning mistakes, particularly mathematical reasoning mistakes. PedCoT consists of pedagogical principles for prompts (PPP) design, two-stage interaction process (TIP) and grounded PedCoT prompts, all inspired by the educational theory of the Bloom Cognitive Model (BCM). We evaluate our approach on two public datasets featuring math problems of varying difficulty levels. The experiments demonstrate that our zero-shot prompting strategy significantly outperforms strong baselines. The proposed method can achieve the goal of reliable mathematical mistake identification and provide a foundation for automatic math answer grading. The results underscore the significance of educational theory, serving as domain knowledge, in guiding prompting strategy design for addressing challenging tasks with LLMs effectively.

Fine-tuning pre-trained large language models (LLMs) with limited hardware presents challenges due to GPU memory constraints. Various distributed fine-tuning methods have been proposed to alleviate memory constraints on GPU. However, determining the most effective method for achieving rapid fine-tuning while preventing GPU out-of-memory issues in a given environment remains unclear. To address this challenge, we introduce LLMem, a solution that estimates the GPU memory consumption when applying distributed fine-tuning methods across multiple GPUs and identifies the optimal method. We conduct GPU memory usage estimation prior to fine-tuning, leveraging the fundamental structure of transformer-based decoder models and the memory usage distribution of each method. Experimental results show that LLMem accurately estimates peak GPU memory usage on a single GPU, with an error rate of up to 1.6%. Additionally, it shows an average error rate of 3.0% when applying distributed fine-tuning methods to LLMs with more than a billion parameters on multi-GPU setups.

GraphQL offers an efficient, powerful, and flexible alternative to REST APIs. However, application developers writing GraphQL clients need both technical and domain-specific expertise to reap its benefits, and avoid over-fetching or under-fetching data. Automated GraphQL generation has so far proven to be a hard problem because of complex GraphQL schema and lack of benchmark datasets. To address these issues, our work focuses on building an LLM-powered pipeline that can accept user requirements in natural language along with the complex GraphQL schema and automatically produce the GraphQL query needed to retrieve the necessary data. Automated GraphQL generation helps reduce entry barriers to application developers, broadening GraphQL adoption.

Knowledge Graphs (KGs) are pivotal in various NLP applications but often grapple with incompleteness, especially due to the long-tail problem where infrequent, unpopular relationships drastically reduce the KG completion performance. In this paper, we focus on Few-shot Knowledge Graph Completion (FKGC), a task addressing these gaps in long-tail scenarios. Amidst the rapid evolution of Large Language Models, we propose a generation-based FKGC paradigm facilitated by LLM distillation. Our MuKDC framework employs multi-level knowledge distillation for few-shot KG completion, generating supplementary knowledge to mitigate data scarcity in few-shot environments. MuKDC comprises two primary components: Multi-level Knowledge Generation, which enriches the KG at various levels, and Consistency Assessment, to ensure the coherence and reliability of the generated knowledge. Most notably, our method achieves SOTA results in both FKGC and multi-modal FKGC benchmarks, significantly advancing KG completion and enhancing the understanding and application of LLMs in structured knowledge generation and assessment.

Most Graph Neural Networks (GNNs) are based on the homophily assumption, where nodes with the same labels or similar features tend to be connected to each other. However, real-world graphs often do not adhere to this homophily assumption. Currently, most researches aggregate multi-hop neighbor information to discover more potentially relevant nodes. However, in the aggregation process of GNNs, the difference in modeling global and local information is not considered, inevitably leading to information loss. Motivated by this limitation, we propose LG-GNN, a local-global adaptive graph neural network for modeling both homophily and heterophily. Specifically, we model the long-range structural similarity and local feature similarity between nodes from global and local perspectives, in order to capture distant dependencies in highly heterophilic networks while reducing the mixing of locally dissimilar feature nodes, thereby increasing the effectiveness of information aggregation in highly heterophilic graphs. Extensive experiments on a wide range of real-world datasets demonstrate that our proposed approach performs well in both heterophilic and homophilic graphs.

Graph anomaly detection (GAD), which aims to identify those graphs that are significantly different from other ones, has gained growing attention in many real-world scenarios. However, existing GAD methods are generally designed for centralized training, while in real-world collaboration, graph data is generally distributed across various clients and exhibits significant non-IID characteristics. To tackle this challenge, we propose a federated graph anomaly detection framework with local-global anomaly awareness (LG-FGAD). We first introduce a self-adversarial generation module and train a discriminator to identify the generated anomalous graphs from the normal graph. To enhance the anomaly awareness of the model, we propose to maximize/minimize the mutual information from local and global perspectives. Importantly, to alleviate the impact of non-IID problems in collaborative learning, we propose a dual knowledge distillation module. The knowledge distillation is conducted over both logits and embedding distributions, and only the student model engages in collaboration to preserve the personalization of each client. Empirical results on various types of real-world datasets prove the superiority of our method.