T01: Advances in Debating Technologies: Building AI That Can Debate Humans
Roy Bar-Haim, Liat Ein-Dor, Matan Orbach, Elad Venezian and Noam Slonim
Engaging in a competitive debate with a human expert, first demonstrated by IBM’s Project Debater in 2019, is a new type of grand challenge that pushes the boundaries of AI. The tutorial addresses the scientific challenges that arise in developing the various components of a debating system, and discusses potential applications of debating technologies.
T02: AI Planning: Theory and Practice
Shirin Sohrabi, Michael Katz and Octavian Udrea
The tutorial provides a theoretical background on AI Planning and introduces some of the existing tools, as well as existing applications that were tackled with these tools. We show how to use the tools on an example application. The participants will get the knowledge and hands-on experience necessary to start using AI Planning tools in their applications.
T03: Bayesian Inference for Deep Learning
Simone Rossi and Maurizio Filippone
Throughout the last decade, the practical advancements and the theoretical understanding of deep learning (DL) models and practices has arguably reached a level of maturity such that it is the preferred choice for any practitioner seeking simple yet powerful solutions to solve machine learning (ML)-related problems. With this tutorial we aim to expose the participants to novel trends in DL for scenarios where quantification of uncertainty matters and we will discuss new and emerging trends in the Bayesian deep learning community.
T04: Cognitive Vision: On Deep Semantics for Explainable Visuospatial Computing
Mehul Bhatt and Jakob Suchan
The tutorial on cognitive vision addresses computational vision and perception at the interface of language, logic, cognition, and artificial intelligence. With a focus on explainable visual sensemaking of dynamic visuospatial imagery, the tutorial demonstrates the integration of methods from AI and Vision with a focus on (combining) declarative reasoning & learning about space, action, motion, and interaction. The tutorial is presented in the applied backdrop of areas as diverse as autonomous driving, cognitive robotics, design (for visual art, architecture, visuoauditory digital media), and behavioural visual perception research in cognitive psychology. The tutorial positions an emerging line of interdisciplinary research bringing together AI, Vision, Psychology, and Design.
T05: Communication Efficient Distributed Learning
Xiaorui Liu, Yao Li, Ming Yan and Jiliang Tang
This tutorial covers the frontiers of communication efficient distributed learning with the focuses on communication compression and decentralization. We will discuss the algorithm developments, theoretical properties and practical implementations.
T06: Complex Event Processing: Languages, Recognition and Forecasting
Elias Alevizos and Alexander Artikis
The recognition and forecasting of events in the multitude of data streams that are being recorded, ranging from business process data to computer and sensor network data, is becoming ever more important. This tutorial shows how formal methods from automata theory and computational logic provide a sound and effective approach to complex event processing in the Big Data era.
T07: Deep Learning for Human Mobility: data, models, and challenges
Luca Pappalardo, Gianni Barlacchi, Massimiliano Luca and Bruno Lepri
Human mobility plays a fundamental role in our society, affecting crucial aspects such as the spreading of viral diseases (e.g., the COVID-19 pandemics), public and private transportation, citizens’ well-being, and the quality of the environment. In this tutorial, we will cover the fundamental concepts of human mobility, which are often overlooked in the field of AI. We will then give an overview of the deep learning approaches employed to predict the next location an individual will visit ( next-location prediction ), forecast crowd flows in urban contexts ( crowd flow prediction ), and generate realistic individual trajectories ( trajectory generation ). In presenting the leading solutions to these problems, we will also discuss the public mobility datasets commonly used and the deep learning techniques usually utilized, providing practical examples on how to train them and output predictions. Finally, we will discuss the main open technical challenges and outline promising future research directions.
T08: Deep Learning for Recommendations: Fundamentals and Advances
Wenqi Fan, Xiangyu Zhao, Dawei Yin and Jiliang Tang
The tutorial aims to provide a comprehensive overview of the recent developments of advanced techniques in deep recommender systems. We will present the background and foundations of Recommender Systems (RecSys), followed by the illustration of three advanced techniques for building RecSys: (1) Graph Neural Networks (GNNs) for Recommendations, (2) Deep Reinforcement Learning (DRL) for Recommendations, and (3) Automated Machine Learning (AutoML) for Recommendations.
T09: Deep Learning on Graphs for Natural Language Processing
Lingfei Wu, Yu Chen, Heng Ji and Bang Liu
This tutorial of Deep Learning on Graphs for Natural Language Processing (DLG4NLP) will cover relevant and interesting topics on applying deep learning on graph techniques to NLP, including automatic graph construction for NLP, graph representation learning for NLP, advanced GNN based models (e.g., graph2seq, graph2tree, and graph2graph) for NLP, and the applications of GNNs in various NLP tasks (e.g., machine translation, natural language generation, information extraction and semantic parsing). In addition, handson demonstration sessions will be included to help the audience gain practical experience on applying GNNs to solve challenging NLP problems using our recently developed open source library – Graph4NLP, the first library for researchers and practitioners for easy use of GNNs for various NLP tasks.
T10: From Statistical Relational to Neural Symbolic Artificial Intelligence
Luc De Raedt, Sebastijan Dumancic, Robin Manhaeve and Giuseppe Marra
The fields of Statistical Relational Artificial Intelligence (StarAI) and Neural Symbolic Artificial Intelligence (NeSy) both tackle the problem of integrating learning and reasoning. This tutorial will discuss seven common dimensions for introducing and categorizing StarAI and NeSy approaches.
T11: Incentivized exploration
How do you incentivize self-interested agents to explore when they prefer to exploit? In contrast with traditional formulations of exploration-exploitation tradeoff, agents control the choice of actions, whereas an algorithm can only issue recommendations.
T12: KR&R Meets Cyber-Physical Systems: Formalization, Behavior, Trustworthiness
Marcello Balduccini, Edward Griffor and Edward Griffor
The aim of the tutorial is to provide an introduction to techniques from Knowledge Represen_x0002_tation and Reasoning (KR&R) that can be used to model and understand Cyber-Physical Systems (CPS). Specifically, we will describe KR&R-based methods for the formalization of the structure and interdependencies among the com_x0002_ponents of CPS, of their emerging behavior, and for reasoning about their properties, such as trustworthiness.
T13: Learning with noisy supervision
Masashi Sugiyama, Tongliang Liu, Bo Han, Quanming Yao and Gang Niu
Noisy data is ubiquitous and harms the performance of most learning algorithms, and sometimes makes existing algorithms break down. This tutorial summarizes the most recent noisy-supervision-tolerant techniques, from the viewpoint of statistical learning, deep learning and their applications in industry.
T14: Mechanism Design without Money: Matching, Facility Locations, and Beyond
Haris Aziz, Hau Chan, Hadi Hosseini and Chenhao Wang
The tutorial aims to introduce audiences to algorithmic mechanism design without money and its applications, for strategic environments when the mechanism designers 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 various classical mechanism design settings (e.g., matching and 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.
T15: Modern Aspects of Big Time Series Forecasting
Jan Gasthaus, Tim Januschowski and Yuyang Wang
The objective of this tutorial is to provide a concise and intuitive overview of the most important methods and tools available for solving large-scale forecasting problems. We review the state of the art in both classical modeling of time series and deep learning for forecasting, as well as discuss the practical aspects of building a large scale forecasting system with case studies.
T16: Neural Machine Reasoning
Truyen Tran, Vuong Le, Hung Le and Thao Minh Le
This tutorial reviews recent advances on dynamic neural networks that aim to reach a deliberative reasoning capability. This goes beyond the current associative pattern matching excelled by deep learning.
T17: Quantum Neural Networks for Speech and Natural Language Processing
Pin-Yu Chen, Yen-Chi Chen, Jun Qi and Huck Yang
This tutorial aims to provide a lecture-style technical overview and hands-on exercise on the recent advances in quantum neural networks and their applications to speech and natural language processing. The scope of this tutorial is to offer entry points and sample codes for machine learning researchers interested in the emerging field of quantum computing and machine learning.
T18: Towards Robust Deep Learning Models: Verification, Falsification, and Rectification
Wenjie Ruan, Elena Botoeva, Xinping Yi and Xiaowei Huang
This tutorial aims to introduce the foundations of assessing robustness of deep learning models, presenting a well-structured review of the state-of-the-art techniques to verify their robustness, to detect their vulnerability by falsification techniques, and finally to improve model robustness through rectification. The tutorial is divided into three main parts dedicated to verification, falsification,
T19: Towards Automated Recommender System
Quanming Yao, Yong Li, Chen Gao, Huan Zhao and Yongqi Zhang
Automated machine learning (AutoML) has emerged as a hot topic with both industrial and academic interests. In this tutorial, based on recent exemplar works, we will introduce backgrounds on AutoML, elaborate on how AutoML can help recommender system (RecSys). Finally, we will throw light upon automated graph representation learning, where AutoML can help design better graph neural networks (GNN) for recommendation tasks.
T20: From Standard Summarization to New Tasks and Beyond: Tasks and Methods of Summarization with Manifold Information
Shen Gao and Rui Yan
Text summarization has gradually become one of the frontiers for the artificial intelligence community, and document summarization is a traditional task in the NLP field. With the increasing number of online services, in many real-world application scenarios, summarization service needs to tackle non-standard input, such as manifold information or structure input. Different from the traditional summarization task, these new tasks incorporate additional manifold information or document structure in the summary generation to further improve the performance of the summarization systems. To facilitate the development of the text summarization filed, we summarize existing researches and give an overview of their task definitions and technical implementations in this tutorial. Specifically, we focus on the recent advances and trends in new summarization tasks, including the summarization with document structure and additional knowledge.
A Quarter of Day
T21: Artificial Intelligence Approaches for COVID-19 and Future Pandemics
This tutorial will highlight the different ways in which Artificial Intelligence and Machine Learning research can help (and have helped) in tackling the COVID-19 pandemic. It will provide an expansive overview of different strands of AI research which have already been done to tackle COVID-19, and it will also present future research challenges (and their real-world implications).
T22: Continual Learning Dialogue Systems – Learning on the Job after Model Deployment
Sahisnu Mazumder and Bing Liu
The tutorial focuses on the research topic of building the nextgeneration dialogue systems that can continuously and interactively learn from end-users during conversation (on-the-job learning after deployment) to become more and more powerful over time. We will provide a background of the topic, and discuss existing techniques and open challenges, which we believe, will shape the future of dialogue systems.
T23: Conversational Recommender Systems
Dietmar Jannach and Li Chen
Fueled by the fast developments in natural language processing and deep learning, we have recently witnessed an increased interest in the development of conversational recommender systems, which support the decision-processes of users in an interactive way. In this tutorial, we review the history, state-of-the-art, recent advances, and open challenges of such systems.
T24: Machine reading comprehension
Zhuosheng Zhang and Hai Zhao
This tutorial presents a comprehensive and comparative review on machine reading comprehension, which aims to train machines with the ability of reading comprehension over realworld data and serves as a major goal of artificial intelligence. The discussion covers the scope of background, development, influence, datasets, typical and state-of-the-art techniques, empirical assessments, and recent trends, with a particular attention on the role of recent advanced pre-trained language models.
T25: Neural Text to Speech Synthesis
Xu Tan and Tao Qin
Neural text to speech (TTS) synthesis aims to synthesize intelligible and natural speech from text based on neural networks, which has been a hot topic in artificial intelligence. Inthis tutorial, we will review the fundamental models of neural TTS and introduce a series of works that aim to push thefrontier of TTS research and cover practical TTS product.
T26: NS4NLP: Neural-Symbolic Modeling for Natural Language Processing
Maria Leonor Pacheco and Dan Goldwasser
Understanding natural language discourse requires both powerful neural representations and symbolic reasoning. This tutorial will discuss the challenges and opportunities of combining the two approachesand introduce a Neural-Symbolic framework for NLP
T27: Recent Advances in Reinforcement Learning for Human-AI Collaboration
Sebastian Tschiatschek and Adish Singla
This tutorial will provide an overview of recent advances in designing reinforcement learning (RL) techniques for facilitating human-AI collaboration. The tutorial will cover the following research directions: (i) designing robust multi-agent RL algorithms, (ii) designing reactive RL algorithms for enabling long-term human-AI collaborations, and (iii) empowering humans to teach RL algorithms in order to steer their behavior.
T28: Reinforcement Learning for Education: Opportunities and Challenges
Adish Singla and Goran Radanovic
This tutorial will provide an overview of the research opportunities and challenges in applying reinforcement learning (RL) methods for improving education (ED). The tutorial will focus on two thrusts: (i) RL→ED, i.e., leveraging recent advances in RL to improve the state of the art technology for ED; (ii) ED→RL, i.e., identifying unique challenges in ED that can help nurture technical innovations and next breakthroughs in RL.
T29: Theoretically Unifying Conceptual Explanation and Generalization of DNNs
This tutorial introduces the speaker’s recent studies, which make the first breakthrough in theoretically unifying two classic directions of explainable AI (XAI), i.e. explaining concepts encoded in a DNN and explaining the generalization power of a DNN. A unified theory to explain both the conceptual explanation and the generalization capacity of a DNN has considerable impacts on explaining various AI tasks in both theory and practice, and will receive broad interests from both novices and `experts.