|Adversarial Machine Learning: On The Deeper Secrets of Deep Learning
|Recent research has found out that deep neural networks behave strangely to slight changes in input. This tutorial will talk about this curious, and yet, still poorly understood behavior, diving into the deeper secrets of deep learning.
|Algorithm Configuration: Challenges, Methods and Perspectives
|Since the peak performance of algorithms in AI strongly depends on the used parameter configurations, automated algorithm configuration tools free user from the tedious and time-consuming task of searching for well-performing configurations. The tutorial on algorithm configuration will cover an overview of the basics of algorithm configuration, as well as recent advances and open challenges in the field.
|Bayesian Optimization for Balancing Metrics in Recommender Systems
|Tuning multiple metrics for a recommendation engine in any large scale platform is an incredibly challenging problem. This tutorial will discuss a framework to solve such problems via tools from Bayesian Optimization.
|Belief and Opinion Dynamics and Aggregation in Multi-Agent Systems
|The tutorial aims to provide a comprehensive overview of models of belief and opinion dynamics and aggregation developed in AI and in the area of multi-agent systems (MAS). These models are based on logic, game-theory and social choice theory.
|Causal Inference and Stable Learning
|In this tutorial, we focus on causal inference and stable learning, aiming to explore causal knowledge from observational data to improve the interpretability and stability of machine learning algorithms.
|Combinatorial Approaches for Data Feature Topic Selection and Summarization
|In this tutorial, we will study the role of submodular functions as combinatorial models and sparsity inducing norms in optimization and learning problems in machine learning, and in particular applications such as data summarization, data selection, active and meta learning, feature selection and rule mining.
|Compressed Communication for Large-scale Distributed Deep Learning
|We survey compressed communication methods for distributed deep learning and discuss the theoretical background, as well as practical deployment on TensorFlow and PyTorch. We also present quantitative comparison of the training speed and model accuracy of compressed communication methods on popular deep neural network models and datasets.
|Compression of Deep Learning Models for NLP
|In this tutorial we will discuss six different types of methods for compression of deep learning models for NLP, and applications of such methods across various NLP tasks. The six types of methods include: pruning, quantization, knowl- edge distillation, parameter sharing, matrix decomposition, and other Transformer based methods.
|Computational Game Theory and Its Applications
|The proposed tutorial aims to introduce audiences to computational game theory and its applications for modeling and solving strategic decision-making processes involving multi-agents in societal domains. The audiences will be exposed to various game-theoretic models, solution concepts, and algorithmic tools, as well as some recent applications of game-theoretic research.
|Current and Future Trends of Neural Knowledge Graph Representation and Reasoning
|In this tutorial, we introduce different deep network architectures that can be trained to perform deductive reasoning with high precision and recall. We will talk about the accuracy, scalability, transferability, generalizability, speed, and interpretability capability of existing and new deep learning approaches and will talk about possible new models to enhance such desirable capabilities.
|Conscious AI: Significance and Development
|Understanding and implementing the ability of imagination by utilizing knowledge that is accumulating in consciousness research is an effective strategy for developing AI with a human-level generalization ability. Three speakers who are conducting research on artificial consciousness and brain architecture based on deep learning will provide a review on the progress of studies related to consciousness, and introduce approaches for the development of artificial consciousness.
|Ethics in Sociotechnical Systems
|The surprising capabilities demonstrated by AI technologies overlaid on detailed data and fine-grained control give cause for concern that agents can wield enormous power over human welfare, drawing increasing attention to ethics in AI. This tutorial introduces ethics as a sociotechnical construct, demonstrating how ethics can be modeled and analyzed, and requirements on ethics (value preferences) can be elicited, in a sociotechnical system.
|Exploring Attention, Dynamic Information Flow, and Modularity as Ingredients for Generalization in Deep Learning
|Deep networks have achieved excellent results but still struggle to match human capabilities in robustness, out-of-domain generalization, and adaptivity. This tutorial explores how new techniques which make neural nets more dynamic and modular help to address these challenges, and will teach researchers practical techniques that they can immediately apply to their own problems.
|Exploring Rare Categories on Graphs: Representation, Inference, and Generalization
|Rare category analysis refers to the problem of studying rare examples from the under-represented minority groups in an imbalanced data, which are observed in a variety of high-impact applications, such as fraud detection, emerging trend detection, rare disease diagnosis, and de-nov drug discovery. This tutorial aims to provide a concise review of the recent developments on rare category characterization, inference, and generalization in the data presented as graphs.
|Fact-Checking, Fake News, Propaganda, and Media Bias: Truth Seeking in the Post-Truth Era
|The tutorial will cover recent work on a number of related problems such as misinformation, disinformation, “fake news”, rumor, and clickbait detection, fact-checking, stance, bias and propaganda detection, source reliability estimation, as well as detecting bots, trolls, and seminar users. We will also discuss recent advances in automatic generation of text, e.g., GPT-2 and GROVER, of images and of videos, e.g., “deep fakes”, and their implication for robojournalism and “fake news” generation.
|Fair AI in a Nutshell: Can Algorithms be Fair?
|In recent years there has been a lot of interest in algorithmic fairness in machine learning; this research area aims to enhance learning algorithms with fairness requirements, namely ensuring that sensitive information (e.g. gender, race, political, and sexual orientation) does not ‘unfairly’ influence the outcome of a learning algorithm. This tutorial aims at describing the state-of-the-art on algorithmic fairness as well as discussing currently unexplored areas of research such as the problems of representation (e.g. learn fair graph embedding) and temporal unfairness (e.g. detect shift in unfair biases over time).
|Federated Learning Systems: Comparative Studies and Hand-on Demonstrations
|The tutorial introduces the concept and taxonomy of federated learning systems and the characteristics of existing federated learning systems from academia and industry. It includes 1) Comparative studies on existing system from the perspective of data partitioning, machine learning model, scale of federation, communication architectures, privacy mechanisms, and motivation of federations, 2) Hand-on demonstrations on how to use existing systems to solve the federated learning use cases in our benchmark.
|Federated Recommender Systems
|Federated recommender systems enable multiple parties to trains recommendation models collaboratively in a secure and privacy-preserving way, without data leaving their own repositories. In this tutorial, we categorize the federated recommender systems, and implement an open-sourced tool that contains typical algorithms for each category, and finally we demonstrate applications in news recommendation and online advertising.
|From Data Independence to Ontology Based Data Access (and back)
|Among the most commonly cited features of the ontology based data access (OBDA) approach to accessing data sources is its ability to use a high-level user-friendly interface to a conceptual understanding of the data (aka ontologies), while still utilizing low-level but efficient ways of representing the data in a computer store. The aim of this tutorial is to compare and contrast this OBDA based approach with approaches centered around the concept of data independence that has been under development in the area of database systems since the early 1970s. The tutorial focuses on the common lessons shared by all approaches, and on how each can benefit from lessons learned from the other.
|Goal Recognition Design
|Goal Recognition Design is the task of redesigning environments in order to facilitate goal recognition, which is the ability to infer agents’ goals from their observed behavior. The tutorial will introduce the topic and demonstrate its importance and relevance to a variety of applications in which efficient goal recognition is essential and in which the environment can be redesigned, including intrusion detection, assisted cognition, computer games, and human-robot collaboration.
|Heterogeneous Information Network Embedding and Applications
|Heterogeneous information network (HIN) embedding has gained considerable attention in recent year, which aims to learn low-dimensional representations of nodes while preserving the semantics and structural properties of HINs. This tutorial will give a survey on recent developments of this field and its application in real applications.
|Logic-Enabled Verification and Explanation of ML Models
|We present an overview of recent trends in verification and explainability of machine learning (ML) models from the formal logic standpoint. Our goal is to illustrate to the AI audience how powerful logic-based methods can be employed to solve a wide range of practical applications, including certifying robustness, safety of ML models, generating explanations of ML models decisions, etc. Therefore, the primary objective of this tutorial is to introduce and explain a topic of emerging importance for the Artificial Intelligence researcher and practitioner.
|Machine Ethics State-of-the art and interdisciplinary challenges
|Much is written and discussed about AI and ethics in both science and media, which can make it confusing to discern where science ends and fad begins. This tutorial serves as a guide through the state-of-the-art and main challenges in machine ethics, contrasting it also with the fields of explainable AI (XAI) and Fairness, Accountability, and Transparency (FAccT).
|Machine Learning and Game Theory
|This tutorial features the recent advances in integrating machine learning with game theory. It covers several topics, including end-to-end learning for strategic decision making, learning-enhanced strategy generation, and adversarial machine learning. It will also cover how these techniques have been used to handle challenges in cyber security, wildlife conservation, and other domains.
|Machine Learning for Combinatorial Optimization
|This tutorial will provide an overview of the recent impact machine learning is having on combinatorial optimization, particularly under the Mixed Integer Programming (MIP) framework. Topics covered will include ML and reinforcement learning for predicting feasible solutions, improving exact solvers with ML, and Ecole, a software framework for learning in exact MIP solvers.
|Machine learning for data streams with scikit-multiflow
|In this tutorial we are going to introduce attendees to data stream mining procedures and examples of big data stream mining applications. Besides the theory we will also present examples using the scikit-multiflow framework, a novel open source Python framework.
|Machine Learning for Drug Development
|In this tutorial, we cover key advancements in machine learning over the last few years, with an emphasis on fundamentally new opportunities in drug development enabled by these advancements.
|Making Your Research Reproducible – Practical Advice on How to Implement the General Guidelines for Making Empirical AI Research Reproducible
|Reproducible research is good research. In this tutorial, we will present the general guidelines for making empirical AI research reproducible and provide practical advice on how to implement them in your research.
|Meta-learning and Automated Machine Learning: Approaches and Applications
|We introduce the most recent updates and advances in meta-learning and AutoML during the past years. And all the slides and our experiences in winning the second place in NeuralIPS 2018 AutoML Competition and ACML 2019 AutoWSL Competition will be shared with all the audiences.
|Mining User Interests from Social Media
|The abundance of user generated content on social media provides the opportunity to build models that are able to accurately and effectively extract, mine and predict users’ interests with the hopes of enabling more effective user engagement, better quality delivery of appropriate services and higher user satisfaction. While traditional methods for building user profiles relied on AI-based preference elicitation techniques that could have been considered to be intrusive and undesirable by the users, more recent advances are focused on a non-intrusive yet accurate way of determining users’ interests and preferences.
|Multimodal Learning in K-12 Education: Promise, Progress and Challenges
|In this tutorial, we will comprehensively review recent developments of applying multimodal learning approaches in AIED, with a focus on those classroom multimodal data. Beyond introducing the recent advances of computer vision, speech, natural language processing in education respectively, we will discuss how to combine data from different modalities and build AI driven educational applications on top of these data.
|The Musical Metacreation tutorial will introduce and present the major AI techniques used to develop generative music systems. The presentation will be illustrated by many historically significant examples.
|Next-Generation Recommender Systems and Their Advanced Applications
|This tutorial presents state-of-the-art theories and approaches to equip the next-generation recommender systems (RS), i.e., the latest and most advanced RS, and their advanced applica- tions. First, we will present the background and foundations of RS, followed by the illustration of three typical theories and approaches for building various next-generation RS: (1) sequential or session-based RS, (2) graph learning based RS, and (3) interactive and conversational RS, together with their prototypes. Finally, we will demonstrate three representative emerging real-world applied cases of RS in fashion industry using artificial intelligence (FashionAI), financial industry using technology (FinTech), and healthcare.
|Optimization & Learning Approaches to Resource Allocation for Social Good
|Communities worldwide face difficult challenges related to disease, poverty, homelessness, and an array of related issues. This tutorial will cover how techniques from optimization and machine learning can be used to improve resource allocation in a range of social good domains, focusing on public health, social work, and healthcare.
|Probabilistic Circuits: Representations, Inference, Learning and Applications
|Exact and efficient probabilistic inference and learning are becoming more and more mandatory when we want to quickly take complex decisions in presence of uncertainty in real-world scenarios where approximations are not a viable option. In this tutorial, we will introduce probabilistic circuits (PCs) as a unified computational framework to represent and learn deep probabilistic models guaranteeing tractable inference. Differently from other deep neural estimators such as variational autoencoders and normalizing flows, PCs enable large classes of tractable inference with little or no compromise in terms of model expressiveness. Moreover, after showing a unified view to learn PCs from data and several real-world applications, we will cast many popular tractable models in the framework of PCs while leveraging it to theoretically trace the boundaries of tractable probabilistic inference.
|Robust Multi-view Visual Learning: A Knowledge Flow Perspective
|Multi-view data are extensively accessible nowadays thanks to various types of features, viewpoints and different sensors. This tutorial covers most multi-view visual data representation approaches from two knowledge flows perspectives, i.e., knowledge fusion and knowledge transfer, centered from conventional multi-view learning to zero-shot learning, and from transfer learning to few-shot learning.
|Rule-based Stream Reasoning
|In many applications of the Internet of Things (IoT), manufacturing digitalization, or cyber-physical systems decisions must be made in a (near-) real-time over continuously changing data. This tutorial provides an in-depth introduction to stream reasoning formalisms with the focus on approaches originating in logic programming.
|Scalable Deep Learning: How far is one billion neurons?
|The tutorial covers the emerging topic of scalable deep learning which makes use of static and adaptive sparse connectivity in neural networks before and throughout training.
|Theoretical Foundations of multi-agent Flexible Temporal Epistemic and Contingent Aspects of Planning
|This tutorial covers the theoretical foundations of multi-agent, flexible, temporal, epistemic and contingent aspects of planning (decentralized partially observable Markov decision processes, dynamic epistemic logic, temporal planning). The tutorial relies on pedagogical tools Hintikka’s world 1 and TouIST2.
|The role of AI in developing persistent personalized privacy and online deception awareness
|The initial optimism about the positive potentials of the Internet and social media has given way to concerns that people are being manipulated through the constant harvesting of personal information and through the control over the information they see online. The goal of this tutorial is (1) to provide the attendees with an understanding of the landscape of online deception, (2) highlight why AI is the perfect vehicle to provide persistent personalized privacy and online deception awareness and (3) detail the possible AI research venues.
|The Role of Knowledge Repositories in Information Retrieval
|As search engines move closer to returning results that are not just a set of links but more directly answer queries, retrieval of relevant results increasingly relies on finding and matching relevant entities within queries and results. The tutorial reviews the role of knowledge repositories in information retrieval, including available choices of knowledge repositories; analyzing their impact and applications in information retrieval; and discussing their strengths and limitations relative to standard retrieval.
|Towards Deep Explanation in Machine Learning Supported by Visual Methods
|This tutorial covers the state-of-the-art research, development, and applications in the area of Interpretable Knowledge Discovery and ML boosted by Visual Methods. The topic is interdisciplinary, bridging efforts of research and applied communities in AI, Machine Learning, Visual Analytics, Information Visualization, and HCI.
|Trusting AI by Testing and Rating Third Party Offerings
|Human’s trust in AI systems is considered as a critical issue in enabling mass-scale adoption of AI technology. In this tutorial, we will make students, developers and researchers aware of the relevant trust issues, focus on AI services offered as third party services without access to code and training data, summarize recently proposed methods to test services, introduce approaches to rate AI services, and familiarize attendees with open source resources.
|Trustworthiness of Interpretable Machine Learning
|This tutorial focuses on the emerging research direction of interpretable machine learning. In particular, the tutorial makes a comprehensive survey on the trustworthiness of current interpretable machine learning methods and the trustworthiness of network features, and introduces several new findings in recent papers of the speakers.
|Tutorial on Robot Audition Open Source Software HARK
|A half-day tutorial for robot audition open source software HARK consisting of lectures, hands-on lessons, and live demonstrations.
|Video-based Data Collection for Sports Tactical Analysis
|Sports match video keeps most information in the games. In this talk, taking badminton as an example, we will introduce how deep learning and machine learning can help in the collection of microscopic data from sports match video. In addition, some interesting results of tactical analysis will be given.