Tutorials

Jingyuan Sun, Weihao Xia, Jixing Li, Shaonan Wang, Jiajun Zhang, A. Cengiz Oztireli, Sien Moens

Topic: Neuro | Planned Length: 1

Abstract:
In recent years, Deep Learning (DL) has made significant progress, with large foundational models showing near-human performance in various tasks across different areas. These advances highlight DL’s potential to help us understand how the human brain processes and interprets these types of information, particularly through brain encoding and decoding. Brain encoding involves mapping stimuli to predict the neural responses they trigger, while brain decoding reconstructs perceived or imagined stimuli from brain activity. DL models, which excel at identifying and manipulating features in data, are well-suited for linking stimuli with brain activities and vice versa. This tutorial will focus on how DL can support brain encoding and decoding, with a particular emphasis on language, visual and speech perception. We will explain the principles and methods of using DL for brain encoding and decoding, presenting both overview of well cited related works and examples of specific studies. We will also discuss the challenges and future directions in this field. By featuring insights from experienced researchers in each area, we aim to provide a thorough and informative overview of the intersection between DL and cognitive neuroscience, encouraging future research, welcoming both junior scholars and experts in this rapidly evolving area.

Da Yan, Lyuheng Yuan, Akhlaque Ahmad, Chenguang Zheng, Hongzhi Chen, James Cheng

Topic: Graphs, DM, ML | Planned Length: 0.25

Abstract:
Graph-theoretic algorithms and graph machine learning models are essential tools for addressing many real-life problems, such as social network analysis and bioinformatics. To support large-scale graph analytics, graph-parallel systems have been actively developed for over one decade, such as Google’s Pregel and Spark’s GraphX, which (i) promote a think-like-a-vertex computing model and target (ii) iterative algorithms and (iii) those problems that output a value for each vertex. However, this model is too restricted for supporting the rich set of heterogeneous operations for graph analytics and machine learning that many real applications demand.

In recent years, two new trends emerge in graph-parallel systems research: (1) a novel think-like-a-task computing model that can efficiently support the various computationally expensive problems of subgraph search; and (2) scalable systems for learning graph neural networks. These systems effectively complement the diversity needs of graph-parallel tools that can flexibly work together in a comprehensive graph processing pipeline for real applications, with the capability of capturing structural features. This tutorial will provide an effective categorization of the recent systems in these two directions based on their computing models and adopted techniques, and will review the key design ideas of these systems.

Haoteng YIN, Beatrice Bevilacqua, Bruno Ribeiro, Pan Li

Topic: Graphs, DM, ML | Planned Length: 0.25

Abstract:
The tutorial examines the foundations and limitations of Graph Neural Networks (GNNs), revealing their bounded effectiveness in distinguishing non-isomorphic graphs and handling prediction tasks involving link and higher-order patterns. First, attendees will explore a novel family of GNNs, offering enhanced expressive power by leveraging the decomposition of graphs into sets of their induced subgraphs. Then, we will demonstrate how these subgraph GNNs can be conceptualized as part of a broader framework of distance encoding schemes, and present their implementation in various tasks involving multiple nodes. Furthermore, attendees will gain an intuitive understanding of how to mitigate the computational challenges posed by learning on subgraphs with structural encodings through a scalable paradigm of algorithm and system co-design. Building upon the introduced theoretical principles, this tutorial presents exciting cutting-edge applications of subgraph-based methods, ranging from biochemistry to programming languages. Through easy-to-follow hands-on coding demonstrations, participants will witness the superior performance and potential of subgraph methods in drug design and neurological disorder diagnosis. The tutorial concludes with future directions and open challenges, emphasizing how this unified framework for subgraph-based representations can help drive future advancements in graph representation learning.

Dacheng Tao, Dongjin Song, Xikun ZHANG

Topic: Graphs, DM, ML | Planned Length: 0.25

Abstract:
In this tutorial, we will introduce this newly emerging area – Continual Graph Learning (CGL). Specifically, we will (1) introduce different CGL settings based on various application scenarios, (2) present the key challenges in CGL, (3) highlight the existing CGL techniques and benchmarks, and (4) discuss potential future directions, as well as the relationship between CGL and various other research directions.

Xin Wang, Haoyang Li, Zeyang Zhang, Wenwu Zhu

Topic: Graphs, DM, ML | Planned Length: 0.25

Abstract:
Graph machine learning has been extensively studied in both academia and industry. Although booming with a vast number of emerging methods and techniques, most of the literature is built on the in-distribution (I.D.) hypothesis, i.e., testing and training graph data are sampled from the identical distribution. However, this I.D. hypothesis can hardly be satisfied in many real-world graph scenarios where the model performance substantially degrades when there exist distribution shifts between testing and training graph data. To solve this critical problem, several advanced graph machine learning techniques which go beyond the I.D. hypothesis, have made great progress and attracted ever-increasing attention from the research community. This tutorial is to disseminate and promote the recent research achievement on graph out-of-distribution adaptation, graph out-of-distribution generalization, and large language models for tackling distribution shifts, which are exciting and fast-growing research directions in the general field of machine learning and data mining. We will advocate novel, high-quality research findings, as well as innovative solutions to the challenging problems in graph machine learning under distribution shifts and the applications on graphs. This topic is at the core of the scope of IJCAI, and is attractive to machine learning as well as data mining audience from both academia and industry.

André Artelt, Ulrike Kuhl, Mark Keane

Topic: XAI | Planned Length: 1 | Link: https://sites.google.com/view/tut-counterfactuals-ijcai24/

Abstract:
Recent research has extensively explored counterfactual explanations, with over a thousand papers proposing at least 150 distinct algorithms, offering a psychologically appealing and regulatory-compliant solution to the eXplainable AI (XAI) problem. This tutorial will provide a comprehensive, practical guide to this topic, with hands-on sessions on theoretical foundations, modeling approaches, and both computational and psychological evaluation methodologies.

Ningyu Zhang, Jia-Chen Gu, Yunzhi Yao, Mengru Wang, Xiang Chen, Shumin Deng

Topic: NLP | Planned Length: 0.5

Abstract:
This tutorial aims to provide a systematic introduction to the latest knowledge editing techniques for large language models, equipping researchers and practitioners from academia and industry with practical tools and methodological overviews. The tutorial will focus on how to efficiently modify and update the behavior and knowledge of large language models in specific domains without compromising their overall performance, addressing limitations in factual accuracy, logical consistency, and harmful content generation.

Bang Liu, Yu Chen, Manling Li, Heng Ji, Lingfei Wu

Topic: NLP | Planned Length: 0.5

Abstract:
The field of AI-generated content has grown rapidly in recent years, thanks to the emergence of large language models and diffusion models that can generate text and images, respectively. To fully explore these developments in the IJCAI community, we propose a tutorial that introduces the foundations and research frontiers of AI-generated content, with a focus on text, image, and video generation. Our tutorial will provide a comprehensive overview of AI-generated content, covering its foundations, frontiers, applications, and societal implications. It will cover the basics of large language models and diffusion models, as well as recent research and applications in this area. We will also discuss the societal concerns surrounding AI-generated content, including AI ethics and safety. By the end of the tutorial, attendees will have a better understanding of the current state of the field and the opportunities and challenges it presents. Our tutorial will be useful for researchers and practitioners interested in the application of AI-generated content to various domains. Attendees will gain insights into the latest techniques and tools for generating high-quality content and learn about the potential benefits and risks associated with this technology.

Sarah Keren, Valentin Robu

Topic: MAS | Planned Length: 0.5

Abstract:
Energy systems are undergoing fundamental changes, with increasing uncertainty, decentralisation, and digitalization, and AI-based solutions are playing an increasingly important role in their management and control. In this tutorial we will look at the fundamental opportunities and challenges in applying AI to this key application domain. The tutorial is aimed at both AI researchers (with a particular focus on PhD students and young researchers) wanting to learn more about challenges in applying AI to this area, as well as practitioners from the energy sector.

Alicia Vidler, Toby Walsh

Topic: ABM | Planned Length: 0.25

Abstract:
The goal of this tutorial is to illustrate and detail an interdisciplinary approach to financial market modeling using agent based systems. Exploring complex adaptive financial trading environments through multi-agent based simulation methods presents an innovative approach within the realm of quantitative finance. We begin with an introduction to financial makers such that a basic knowledge of agent based modeling is assumed but no prior technical knowledge of financial markets are required.

Despite the dominance of multi-agent reinforcement learning approaches in financial markets with observable data, there exists a set of systematically significant financial markets that pose challenges due to their partial or obscured data availability. We will explore the techniques, challenges and requirements needed to apply agent based modeling methods to such systematically important bond markets. The tutorial will also necessarily discuss features and aspects of what makes bond markets both unique, and why developing AI and, in particular, explainable AI methods, for markets have application to other areas of finance. This tutorial is designed to be of interest to a wide range of audience. The inter-disciplinary approach will allow a technical focus to be of special interest to industry practitioners whilst researches will find details around financial market design and applications insightful.

Hau Chan, Minming Li

Topic: OPT, GTEP | Planned Length: 0.25

Abstract:
This tutorial aims to introduce audiences to the paradigm of mechanism design without money for facility location problems and their potential real-world applications. In such a paradigm, the mechanism designers (or social planners) are required to elicit private information from the agents in order to generate desirable outcomes and implement desirable mechanisms’ properties when monetary transfers are not allowed. The audiences will be exposed to classical mechanism design settings of facility locations, mechanisms’ desired properties and solution concepts, and algorithmic tools/mechanisms. The tutorial will also cover some recent directions and applications of mechanism design without money for facility location problems.

Albert Bifet, Bernhard Pfahringer, Heitor Gomes, Nuwan Gunasekara

Topic: ML | Planned Length: 0.5

Abstract:
Machine Learning for Data Streams (MLDS) has been an important area of research since the late 1990s, and its usage in industry has grown significantly over the last few years. However, there is still a gap between the cutting-edge research and the tools that are readily available, which makes it challenging for practitioners, including experienced data scientists, to implement and evaluate these methods in this highly complex domain. Our tutorial aims to bridge this gap with a dual focus. We discuss advanced research topics, such as partially delayed labeled streams, while providing practical demonstrations of their implementation and assessment using Python. By catering to both researchers and practitioners, the tutorial aims to empower them to design and conduct experiments, and extend existing methodologies.

Xuhui Fan, Zhangkai Wu, Hui Chen, Feng Zhou, Christopher Quinn, Longbing Cao

Topic: ML | Planned Length: 0.5

Abstract:
Deep Variational Learning~(DVL) is proposed as a core foundation in the area of generative AI, providing systematic learning strategies to model uncertainties, enhance robustness, and enable more informed decision-making process. This tutorial will provide an overview of deep variational learning, and focus on its aspects of challenges, theories, algorithms, applications and opportunities.

Soyeon Han, Yihao Ding, Josiah Poon, Seong-Bae Park, Prasenjit Mitra

Topic: NLP | Planned Length: 0.5

Abstract:
Visually Rich Document(VRD) understanding and information extraction include different tasks to understand a document and extract valuable information automatically. Recently, there has been a rising demand for developing document understanding among different domains, including business, law and medicine, in order to boost the efficiency of work associated with many documents. This tutorial introduces the different tasks of document understanding and the recent techniques. Specifically, the tutorial will be presented in a half-day hybrid lecture-laboratory style, first introducing the typical document understanding tasks, including layout analysis, structure parsing, key information extraction and document question answering. It will then explain the recent deep learning-based techniques in document understanding, including feature representations, cross-modality fusing, encoding methods, and pretraining mechanisms. Two Hands-on Laboratories will ultimately show how to apply state-of-the-art document understanding models to tackle distinct downstream tasks, and the industrial demo with the Bank of Korea team will share the service of VRD understanding with LLM in the real world.

Wenqi FAN, Yujuan Ding, Shijie WANG, Liangbo Ning, Qiaoyu Tan, Qing Li

Topic: REC, NLP | Planned Length: 0.5 | Link: https://advanced-recommender-systems.github.io/LLMs4Rec-IJCAI/

Abstract:
Given the explosive growth of information available, recommender systems have become one of the essential services for online activities that effectively assist users in finding the content they are interested in or the product they target to purchase, thereby enhance user engaagement and satisfaction. While recommendation methods have experienced great development boosted by deep learning advances over the past few years, they are still facing several limitations. Existing methods may have difficulties in effectively understanding and capturing textual side information, inabilities in generalizing to various recommendation scenarios and reasoning on their predictions. Recently, the emergence of Large Language Models (LLMs) has revolutionized the field of Natural Language Processing (NLP) for their spectacular language understanding and generation abilities. More impressively, they are capable of reasoning to tackle complex problems and can be easily generalized for new domains or tasks. Such capabilities provide opportunities to address the aforementioned limitations in existing recommendation methods, which makes LLM-powered recommender systems a promising research area in the future. To this end, in this tutorial, we aim to provide a comprehensive overview of the recent progress of LLM-powered recommender systems from various aspects, including Pre-training, Fine-tuning, and Prompting. It is expected to help researchers and industrial practitioners in related fields in the AI community to gain an overview understanding of LLM-powered recommender systems and inspire them with new ideas for more exciting studies in the future.

Muqun (Rachel) Li, Riyaaz Shaik, Fabian Boemer, Karl Tarbe, Haris Mughees, Yuantao Peng, Pranav Prashant Thombre, Wei Xu, Sudhanshu Mohan, Rahul Nim, Rehan Rishi, Noyan Tokgozoglu, Kranthi Chalasani, Chandru Venkataraman

Topic: IR, SEC | Planned Length: 0.5 | Link: https://sites.google.com/view/pirtutorialijcai2024/home

Abstract:
Personalization involves tailoring services or content to individual users based on their preferences, behaviors, and characteristics with machine learning (ML). Balancing personalization with privacy is a critical challenge. Private Information Retrieval (PIR) protocols allow users to access specific data from a server without revealing the details of their queries, which addresses concerns associated with confidentiality in information retrieval scenarios. Homomorphic Encryption (HE), on the other hand, enables computations on encrypted data, preserving the privacy of sensitive information during processing. PIR and HE can be integrated into recommendation systems powered by ML models to enable personalized services while preserving the confidentiality of user data. This intersection provides a privacy-enhanced approach to deliver customized experiences in various domains.

Practical applications and recent advancements in PIR and HE have shown considerable promise in addressing privacy concerns for personalization, but to date these methods have had limited exposure in the ML community.
This tutorial aims to explore the synergies between Personalization, PIR, and HE to create a comprehensive framework for developing privacy-preserving personalized services that could benefit a broad audience that are interested in the theoretical and/or the practical aspects of privacy-preserving machine learning (PPML) with HE.

Kevin Baum, Louise Dennis, Marija Slavkovik

Topic: ETH | Planned Length: 0.5 | Link: https://machineethics.github.io/

Abstract:
Machine ethics is a multi-disciplinary effort concerned with enabling ethical and value-aligned behaviour from machines towards people and other machines. The tutorial will present the state of the art in machine ethics with the goal of enabling researchers to enter this field or reflect critically on the ethical dimensions of their own work.

Ulle Endriss

Topic: GTEP | Planned Length: 0.5 | Link: https://staff.science.uva.nl/u.endriss/teaching/ijcai-2024/

Abstract:
One of the most exciting developments in the field of computational social choice in recent years has been the use of SAT solvers to automate the task of proving theorems regarding the design of methods for multiagent decision making. This tutorial will offer a hands-on introduction to this powerful new approach.

Warut Suksompong

Topic: GTEP | Planned Length: 0.25

Abstract:
Tournaments are commonly used to select winning alternatives in scenarios involving pairwise comparisons such as sports competitions and political elections. In this tutorial, we will survey developments in two active lines of work—tournament solutions and single-elimination tournaments—with a focus on how computational social choice has brought new frameworks and perspectives into these decades-old studies.

Adish Singla

Topic: Education | Planned Length: 0.25

Abstract:
This tutorial will provide an overview of the research opportunities and challenges in applying generative AI methods for improving education. The tutorial will focus on two thrusts: (i) GAI->ED, i.e., leveraging recent advances in generative AI to improve state-of-the-art educational technology; (ii) ED->GAI, i.e., identifying unique challenges in education that require safeguards along with technical innovations in generative AI.

Jiangchao Yao, Feng Liu, Bo Han

Topic: ML, SEC | Planned Length: 0.5

Abstract:
“Knowledge should not be accessible only to those who can pay” said Robert May, chair of UC’s faculty Academic Senate. Similarly, machine learning should not be accessible only to those who can pay. Thus, machine learning should benefit to the whole world, especially for developing countries. When dataset sizes grow bigger, it is laborious and expensive to obtain perfect data (e.g., clean, safe, and balanced data), especially for developing countries. As a result, the volume of imperfect data becomes enormous, e.g., web-scale image and speech data with noisy labels, images with specific noise, and long-tail-distributed data. However, standard machine learning assumes that the supervised information is fully correct and intact. Therefore, imperfect data harms the performance of most of the standard learning algorithms, and sometimes even makes existing algorithms break down. In this tutorial, we focus on trustworthy learning when facing three types of imperfect data: noisy data, adversarial data, and long-tailed data.

Nir Rosenfeld

Topic: ML, SEC, ETH | Planned Length: 0.5

Abstract:
The success of machine learning across a wide array of tasks and applications has made it appealing to use it also in the social domain. Indeed, learned models now form the backbone of recommendation systems, social media platforms, online markets, and e-commerce services, where they are routinely used to inform decisions by, for, and about their human users. But humans are not your conventional input – they have goals, beliefs, and aspirations, and take action to promote their own interests. Given that standard learning methods are not designed to handle inputs that ‘behave’, a natural question is: how should we design learning systems when we know they will be deployed and used in social settings?
This tutorial introduces strategic machine learning, a new and emerging subfield of machine learning that aims to develop a disciplined framework for learning under strategic user behavior. The working hypothesis of strategic ML is simple: users want things, and act to achieve them. Surprisingly, this basic truism is difficult to address within the conventional learning framework. The key challenge is that how users behave often depends on the learned decision rule itself; thus, strategic learning seeks to devise methods which are able to anticipate and accommodate such responsive behavior. Towards this, strategic ML offers a formalism for reasoning about strategic responses, for designing appropriate learning objectives, and for developing practical tools for learning in strategic environments. The tutorial will survey recent and ongoing work in this new domain, present key theoretical and empirical results, provide practical tools, and discuss open questions and landmark challenges.

Xiaoyuan Yi, Jieyu Zhao, Xing Xie

Topic: ETH | Planned Length: 0.5

Abstract:
In the ever-evolving field of AI, big models emerge as essential milestones and revolutionize the role of AI, driving innovation and diverse applications. Yet, as these models become increasingly integrated into human daily life, they also raise potential safety and ethical concerns, necessitating focused research on value alignment technologies. Drawing on comprehensive surveys, this tutorial delves into big model alignment, a pivotal technology to ensure responsible AI development from five aspects: its history tracing back to the 1920s (where it comes from), mathematical essence originating in Reinforcement Learning (what it is), recent advances in alignment goals (what to align) and alignment approaches (how to align), as well as modern challenges of creating optimally aligned AI systems (where to go).
The tutorial part will be organized from two perspectives: (1) Interdisciplinary discussions on alignment goals and their evaluation methods arising from the specification problem. We critically examine the ongoing debate over alignment objectives, from adherence to human instructions, alignment with human preferences, to integration of human values, charting the shift towards a value-driven AI development paradigm. (2) Technical introduction of alignment approaches covering three lines, Reinforcement Learning, Supervised Fine-Tuning, and In-context Learning, as well as emerging concepts such as personal and multimodal alignment. We will detail the mathematical essence of each, their underlying connections, strengths, limitations, and the challenges. By the conclusion of the tutorial, participants will gain an in-depth understanding of the crucial objectives and prospects of value alignment, providing a comprehensive insight into the field’s present status and future trajectory.

Florian Strub, Olivier Pietquin

Topic: NLP, RL | Planned Length: 0.5

Abstract:
While Reinforcement Learning (RL) recently became essential to Large Language Models (LLMs) alignment, we still only scraped the surface of the potential impact of RL on LLMs. Beyond alignment to human preferences, RL genuinely trains LLMs to generate full completions from prompts, potentially outperforming standard supervised learning approaches based on next-token prediction. Contrary to popular belief, the structural properties of the language domain make applying RL a straightforward process. This tutorial thus aims to pedagogically dive into several RL-inspired methods to train language models efficiently. Taking an inductive approach, we use a summarization task as a support to demystify RL-based training: detailing underlying hypotheses underneath online RL(HF) and DPO-like algorithms, hinting at good practices and pitfalls before exploring original approaches such as language sequence modeling and self-play. We expect to democratize the usage of RL in the LLM community and intuite the emergence of new language modeling training paradigms.

Pratik Gajane, Mykola Pechenizkiy, Yingqian Zhang

Topic: RL, ETH | Planned Length: 0.5 | Link: https://fair-rl.github.io/

Abstract:
The tutorial offers a comprehensive overview of fairness in reinforcement learning (RL), spanning various applications and theoretical frameworks. We will start by delving into the crucial issue of fairness in algorithmic decision-making using the paradigm of reinforcement learning across various high-impact real-world domains. We will provide a comparative analysis of the fairness conceptualizations in the (relatively mature) field of fair-ML with the emerging field of fair-RL. Moving forward, we will introduce fundamental concepts of reinforcement learning, including problem settings and performance measures for both utility and fairness, while also reviewing mathematical results and empirical works on fairness-aware (deep) RL solutions. Furthermore, we will explore the considered fairness notions, discuss integrating fairness into reinforcement learning as a multi-objective optimization problem, and examine the trade-off between fairness and utility across various applications. The tutorial will conclude with a discussion of future challenges in fair reinforcement learning and their prospective solutions, exploring interdisciplinary insights from successful fair machine learning research and applying them to the realm of fair reinforcement learning.

Yizhou Zhang, Yan Liu, Lun Du, Karishma Sharma

Topic: SEC, NLP | Planned Length: 0.25

Abstract:
The consistent abuse of misinformation to manipulate public opinion on social media has become increasingly evident in various domains, encompassing politics, as seen in presidential elections, and healthcare, most notably during the recent COVID-19 pandemic. This threat has grown in severity as the development of Large Language Models (LLMs) empowers manipulators to generate highly convincing deceptive content with greater efficiency. Furthermore, the recent strides in chatbots integrated with LLMs, such as ChatGPT, have enabled the creation of human-like interactive social bots, posing a significant challenge to both human users and the social-bot-detection systems of social media platforms.These challenges motivate researchers to develop algorithms to mitigate misinformation and social media manipulations. This tutorial introduces the advanced machine learning researches that are helpful for this goal, including (1) machine learning for modeling social manipulation and misinformation dissemination behaviors, (2) LLM-generated misinformation, and (3) LLM-based misinformation detection. In addition, we also present possible future directions.

Qiongkai Xu, Xuanli He

Topic: NLP | Planned Length: 0.25

Abstract:
In the rapidly evolving landscape of artificial intelligence, Large Language Models (LLMs) have emerged as the cornerstone of recent advancements, captivating the academic and industry sectors with their remarkable capabilities. These models have significantly improved various applications, from natural language processing to content creation. However, the widespread adoption of LLMs has led to notable concerns of their misuse, including (1) plagiarism in education and academic, (2) dissemination of misinformation, and (3) model extraction attacks, threatening the integrity of digital content and intellectual property. This tutorial delves into the heart of these issues, offering an overview of state-of-the-art technologies designed to authenticate the authorship of content generated by LLMs. By exploring these countermeasures, participants will gain valuable insights into safeguarding their work and ensuring the responsible use of these powerful models.

Zhihong Deng, Jing Jiang, Chengqi Zhang

Topic: RL | Planned Length: 0.25

Abstract:
While autonomous agents driven by reinforcement learning techniques have made significant strides in decision-making under uncertainty, they still lack a nuanced understanding of the world. Recognizing the pivotal role of causal understanding in human cognition, a new wave of reinforcement learning research has emerged, integrating insights from causality research to improve agent’s decision-making capabilities. Our tutorial delves into this fascinating research area, offering participants a unique opportunity to explore the intersection of causality and reinforcement learning.
The tutorial is divided into two parts, crafted to provide participants with a comprehensive understanding of causal reinforcement learning. In the first part, participants will be guided through the fundamental concepts underpinning both causality and reinforcement learning, including basic definitions, mathematical representations, and their connections. Moving into the second part, participants will embark on a journey to explore the synergies between causal research and reinforcement learning, gaining insights into how causal research enriches traditional reinforcement learning methodologies, addressing critical challenges such as sample efficiency, generalization ability, spurious correlations, and the integration of explainability, fairness, and safety principles. The tutorial will conclude with a discussion of the future opportunities and challenges in causal reinforcement learning. Whether you are an experienced reinforcement learning researcher, a causal research enthusiast, or a machine learning practitioner eager to expand your toolkit, join us in exploring the exciting world of causal reinforcement learning and advancing this emerging field together.

Ming Jin, Shangding Gu

Topic: RL, SEC | Planned Length: 0.25

Abstract:
Reinforcement learning (RL) offers powerful approaches for developing sophisticated behaviors and decision-making capabilities. However, for high-stakes applications in domains like healthcare, transportation, power, and finance, ensuring safety and reliability is imperative. Importantly, safe RL also holds promise for aligning emergent AI systems like large language models trained online via human feedback. The tutorial will discuss such applications and their unique safety needs. By equipping attendees with conceptual and technical knowledge coupled with an outlook of open problems and interdisciplinary perspectives, this tutorial seeks to align community efforts towards translating cutting-edge safe RL breakthroughs into trustworthy autonomous systems that can be deployed safely with high confidence.

Ana Ozaki

Topic: ETH | Planned Length: 0.25

Abstract:
A major concern when dealing with complex machine learning models, such as language models, is to determine what influences their outcome. This tutorial casts light on Angluin’s exact learning framework and Valiant’s probably approximately correct framework and whether/how they can be employed to systematically probe machine learning models, extracting high level abstractions which can inform about their knowledge, general behaviour, and potentially harmful biases.

Xin Wang, Yuwei Zhou, Hong Chen, Wenwu Zhu

Topic: ML | Planned Length: 0.25

Abstract:
This tutorial focuses on curriculum learning (CL), an important topic in machine learning, which gains an increasing amount of attention in the research community. CL is a learning paradigm that enables machines to learn from easy data to hard data, imitating the meaningful procedure of human learning with curricula. As an easy-to-use plug-in, CL has demonstrated its power in improving the generalization capacity and convergence rate of various models in a wide range of scenarios such as computer vision, natural language processing, data mining, reinforcement learning, etc.
Therefore, it is essential introducing CL to more scholars and researchers in the machine learning community. However, there have been no tutorials on CL so far, motivating the organization of our tutorial on CL at IJCAI 2024.
We will organize the tutorial from the following aspects: (1) theories, (2) approaches, (3) applications, (4) tools and (5) future directions. First, we introduce the motivations, theories and insights behind CL. Second, we advocate novel, high-quality approaches, as well as innovative solutions to the challenging problems in CL. Then we present the applications of CL in various scenarios, followed by some relevant tools. In the end, we discuss open questions and the future direction in the era of large language models. We believe this topic is at the core of the scope of IJCAI and is attractive to the audience interested in machine learning from both academia and industry.