Embedded Files
Real World Applications of Automated Negotiation Technologies Invited Talk 1 on Dec 13th
Real World Applications of Automated Negotiation Technologies Invited Talk 1 on Dec 13th
Satoshi Morinaga, Data Science Research Labs, NEC Corporation
Automatic negotiation technology is a technology that searches for agreeable terms by all negotiation participants through exchanging messages among them. Typically, the search is carried out by repeating a procedure in which one agent proposes an agreement and the other agents answer whether they agree or reject it.
Automatic negotiation technology is a technology that searches for agreeable terms by all negotiation participants through exchanging messages among them. Typically, the search is carried out by repeating a procedure in which one agent proposes an agreement and the other agents answer whether they agree or reject it.
Recently, this technology has begun to be used in real business situations. In this talk, I will introduce two examples of real-world applications of the automated negotiation technology we are developing.
Recently, this technology has begun to be used in real business situations. In this talk, I will introduce two examples of real-world applications of the automated negotiation technology we are developing.
The first is the negotiation of flight plans among drones operated by different economic entities. In many use cases in the real world, it is difficult to resolve the interference between flight plans while satisfying the objectives of the individual operators by defining coordination rules in advance or by a centralized control tower. We will introduce an example of resolving interference through automatic negotiation between operating entities in such use cases.
The first is the negotiation of flight plans among drones operated by different economic entities. In many use cases in the real world, it is difficult to resolve the interference between flight plans while satisfying the objectives of the individual operators by defining coordination rules in advance or by a centralized control tower. We will introduce an example of resolving interference through automatic negotiation between operating entities in such use cases.
The second is the negotiation of commercial terms and conditions between companies. In the supply chain of any industry, such as manufacturing and logistics, there is a large amount of work to find mutually agreeable transaction terms between buyers and sellers of products and services. We present an example of streamlining business negotiations through automatic negotiation between companies, or through automatic negotiation bot technology where a negotiation agent is implemented in only one side of the companies and a human negotiates against the agent in the other side company.
The second is the negotiation of commercial terms and conditions between companies. In the supply chain of any industry, such as manufacturing and logistics, there is a large amount of work to find mutually agreeable transaction terms between buyers and sellers of products and services. We present an example of streamlining business negotiations through automatic negotiation between companies, or through automatic negotiation bot technology where a negotiation agent is implemented in only one side of the companies and a human negotiates against the agent in the other side company.
In both examples, for social implementation, it is necessary to solve not only technical issues but also non-technical issues such as standardization of messages and consistency with business practices. In this talk, I would like to discuss these issues as well.
In both examples, for social implementation, it is necessary to solve not only technical issues but also non-technical issues such as standardization of messages and consistency with business practices. In this talk, I would like to discuss these issues as well.
Satoshi MORINAGA obtained the B.E., M.E., and Ph.D. degrees from the University of Tokyo in 1992, 1994 and 1999, respectively. He joined NEC Corporation in 1994, and was seconded to the Financial Supervisory Agency (FSA) from 2000 to 2008 as a deputy director and a special researcher. Currently, he is an Executive Research Fellow of Data Science Research Laboratories in NEC Corporation, Deputy Director of NEC-AIST AI Cooperative Research Laboratory in AIST, Deputy Director of RIKEN AIP-NEC Collaboration Center in RIKEN, and Chief Digital Officer of BIRD INITIATIVE Inc. In IEICE, he is/was a member of the editorial board of journals, and a member of the technical committee on Information-Based Induction Sciences and Machine Learning. Currently, he is mainly engaged in research and social implementation of mathematical modeling, machine learning, and automated negotiation. He is the author of "The Challenge of Advanced Operational Risk Management" published by Kinzai Institute for Financial Affairs.
Somebody-knows and other notions of group knowledge and belief Invited Talk 2 on Dec 13th
Somebody-knows and other notions of group knowledge and belief Invited Talk 2 on Dec 13th
Thomas Ågotnes, University of Bergen
In the talk I take a critical look at standard notions of group knowledge and belief in multi-agent systems, including common knowledge, distributed knowledge and general knowledge (everybody-knows) as formalised in epistemic logic. The properties of these notions vary depending on the properties we assume of (individual) knowledge, but I will (provocatively) make the argument that in many - if not most - cases, group knowledge (or belief) does not exist. I will particularly focus on a notion of group knowledge that so far has received far less attention than those just mentioned: somebody-knows. While something is general knowledge if it is known by *everyone*, this notion holds if it is known by *someone*. Somebody-knows is thus weaker than general knowledge but stronger than distributed knowledge. I will introduce an logical operator for somebody-knows in the style of standard group knowledge operators, and study its properties. I provide an equivalent neighbourhood semantics for the language with a single somebody-knows operator, together with a completeness result: the somebody-knows operators are completely characterised by the modal logic EMN extended with a particular weak conjunctive closure axiom. The neighbourhood semantics and the completeness and complexity results also carry over to logics for so-called local reasoning (Fagin et al., 1995) with bounded ``frames of mind'', correcting an existing completeness result in the literature (Allen 2005). The talk is based on joint work with Yi N. Wang.
In the talk I take a critical look at standard notions of group knowledge and belief in multi-agent systems, including common knowledge, distributed knowledge and general knowledge (everybody-knows) as formalised in epistemic logic. The properties of these notions vary depending on the properties we assume of (individual) knowledge, but I will (provocatively) make the argument that in many - if not most - cases, group knowledge (or belief) does not exist. I will particularly focus on a notion of group knowledge that so far has received far less attention than those just mentioned: somebody-knows. While something is general knowledge if it is known by *everyone*, this notion holds if it is known by *someone*. Somebody-knows is thus weaker than general knowledge but stronger than distributed knowledge. I will introduce an logical operator for somebody-knows in the style of standard group knowledge operators, and study its properties. I provide an equivalent neighbourhood semantics for the language with a single somebody-knows operator, together with a completeness result: the somebody-knows operators are completely characterised by the modal logic EMN extended with a particular weak conjunctive closure axiom. The neighbourhood semantics and the completeness and complexity results also carry over to logics for so-called local reasoning (Fagin et al., 1995) with bounded ``frames of mind'', correcting an existing completeness result in the literature (Allen 2005). The talk is based on joint work with Yi N. Wang.
Thomas Ågotnes is a full Professor of Information Science at the University of Bergen in Norway, where he is the head the Logic and AI (LAI) research group and currently head of teaching in Information Science. He is also a Changjiang Professor of Logic at the Institute of Logic and Intelligence at Southwest University in Chongqing, People's Republic of China.
Professor Ågotnes' research is in the intersection of formal logic, artificial intelligence and multi-agent systems. He is particularly interested in formalising reasoning about information (ex)change and different aspects of interaction in multi-agent systems, using modal logic. His work is often interdisiplinary, combining formal logic and computer science with formal/mathematical frameworks for modeling interaction from the social sciences such as game theory, social choice theory or social network analysis.
He has published extensively in these areas. See the Research Topics tab for more details about some research topics and collaborations, and the Publications tab for a list of publications. He has (with co-authors) received the best paper award at the AAMAS conference (2009). In 2018 he was awarded the Changjiang (Yangtze River) Scholar Award by the Chinese Ministry of Education. He has served the research community through numerous program committees, including senior program committees of IJCAI and AAMAS, as well as a frequent referee for journals and research councils. He is currently co-chair of the 2020 Scandinavian Logic Symposium (now moved to 2021 due to covid-19), and tutorials co-chair of AAMAS 2021.
Machine Learning in EEG-based Brain-Computer Interfaces Invited Talk 3 on Dec 14th
Machine Learning in EEG-based Brain-Computer Interfaces Invited Talk 3 on Dec 14th
Dongrui Wu, Huazhong University of Science and Technology
A brain-computer interface (BCI) enables a user to communicate with a computer directly using brain signals. Electroencephalogram (EEG) is the most frequently used input signal in non-invasive BCIs. This talk will introduce several newly proposed machine learning approaches for accurate, secure and privacy-preserving EEG-based BCIs.
A brain-computer interface (BCI) enables a user to communicate with a computer directly using brain signals. Electroencephalogram (EEG) is the most frequently used input signal in non-invasive BCIs. This talk will introduce several newly proposed machine learning approaches for accurate, secure and privacy-preserving EEG-based BCIs.
Dongrui Wu received a B.E in Automatic Control from the University of Science and Technology of China, Hefei, China, in 2003, an M.Eng in Electrical and Computer Engineering from the National University of Singapore in 2006, and a PhD in Electrical Engineering from the University of Southern California, Los Angeles, CA, in 2009. He is now Professor and Deputy Director of the Key Laboratory of the Ministry of Education for Image Processing and Intelligent Control, School of Artificial Intelligence and Automation, Huazhong University of Science and Technology, Wuhan, China.
Prof. Wu's research interests include affective computing, brain-computer interface, computational intelligence, and machine learning. He has more than 160 publications (7,600+ Google Scholar citations; h=43), including a book ``Perceptual Computing" (Wiley-IEEE Press, 2010), and 10 patents. He received the IEEE Computational Intelligence Society (CIS) Outstanding PhD Dissertation Award in 2012, the IEEE Transactions on Fuzzy Systems Outstanding Paper Award in 2014, the North American Fuzzy Information Processing Society (NAFIPS) Early Career Award in 2014, the IEEE Systems, Man and Cybernetics (SMC) Society Early Career Award in 2017, the IEEE SMC Society Best Associate Editor Award in 2018, the USERN Prize in Formal Sciences in 2020, and the IEEE International Conference on Mechatronics and Automation Best Paper Award in 2020. He was a selected participant of the Heidelberg Laureate Forum in 2013, the US National Academies Keck Futures Initiative (NAKFI) in 2015, and the US National Academy of Engineering German-American Frontiers of Engineering (GAFOE) in 2015. His team won the First Prize of the China Brain-Computer Interface Competition twice (2019 and 2020).
Prof. Wu is Editor-in-Chief of the IEEE SMC eNewsLetter, and an Associate Editor of the IEEE Transactions on Fuzzy Systems (2011-2018; 2020-), the IEEE Transactions on Human-Machine Systems (since 2014), the IEEE Computational Intelligence Magazine (since 2017), and the IEEE Transactions on Neural Systems and Rehabilitation Engineering (since 2019). He is the Associate Vice-President for Human-Machine Systems of the IEEE SMC Society.
Causality-Inspired Recommender Systems Invited Talk 4 on Dec 14th
Causality-Inspired Recommender Systems Invited Talk 4 on Dec 14th
Guandong Xu, University of Technology Sydney
Causal learning has attracted a lot of research attention with the advance in explainable artificial intelligence. Causal learning contains causal discovery and causal inference two directions, where causal inference is to estimate the causal effects in treatment guided by causal graph structure and has been extended in tasks of counterfactual explanation, counterfactual fairness, disentanglement learning, interpretability, and debiasing. In this talk, we will introduce our latest research progress of incorporating causal learning into recommender systems, and present three recent studies on de-biasing confounding in recommendation, causal disentanglement for Intent Learning in Recommendation, and off-policy learning in recommendation. Experimental studies on real world datasets have proven the effectiveness of the proposed models.
Causal learning has attracted a lot of research attention with the advance in explainable artificial intelligence. Causal learning contains causal discovery and causal inference two directions, where causal inference is to estimate the causal effects in treatment guided by causal graph structure and has been extended in tasks of counterfactual explanation, counterfactual fairness, disentanglement learning, interpretability, and debiasing. In this talk, we will introduce our latest research progress of incorporating causal learning into recommender systems, and present three recent studies on de-biasing confounding in recommendation, causal disentanglement for Intent Learning in Recommendation, and off-policy learning in recommendation. Experimental studies on real world datasets have proven the effectiveness of the proposed models.
Dr Guandong Xu is the Professor at School of Computer Science, University of Technology Sydney, specialising in Data Science, Recommender Systems, and Social Computing. He has published 220+ papers in leading AI and Data Science journals and conferences. He leads Smart Future Research Centre and Data Science and Machine Intelligence Lab at UTS. He is the Editor-in-Chief of Human-centric Intelligent Systems and assistant Editor-in-Chief of World Wide Web Journal and serving in editorial board or guest editors for several international journals. He has received several Awards from academia and industry, e.g., European Finance Management Association Insurance innovation award, Top-10 Australian Analytics Leader and Australian Computer Society Disruptors Award. He holds the Australian Computer Society (ACS) Fellow.
A Principle-based Analysis of Abstract Agent Argumentation Semantics Invited Talk 5 on Dec 14th
A Principle-based Analysis of Abstract Agent Argumentation Semantics Invited Talk 5 on Dec 14th
Liuwen Yu, University of Luxembourg
Abstract agent argumentation frameworks extend Dung’s theory with agents, and in this paper we study four types of semantics for them. First, agent defense semantics replaces Dung’s notion of defense by some kind of agent defense. Second, social agent semantics prefers arguments that belong to more agents. Third, agent reduction semantics considers the perspective of individual agents. Fourth, agent filtering semantics are inspired by a lack of knowledge. We study five existing principles and we introduce twelve new ones. In total, we provide a full analysis of fifty-two agent semantics and seventeen principles.
Abstract agent argumentation frameworks extend Dung’s theory with agents, and in this paper we study four types of semantics for them. First, agent defense semantics replaces Dung’s notion of defense by some kind of agent defense. Second, social agent semantics prefers arguments that belong to more agents. Third, agent reduction semantics considers the perspective of individual agents. Fourth, agent filtering semantics are inspired by a lack of knowledge. We study five existing principles and we introduce twelve new ones. In total, we provide a full analysis of fifty-two agent semantics and seventeen principles.
Liuwen Yu is a doctoral researcher in the Law, Science and Technology Joint Doctorate programme funded by Marie Skłodowska-Curie Actions, supervised by Prof. Leon Van der Torre. She received her master degree in Logic from Zhejiang University (China) in 2018, with a dissertation in Argumentation entitled " The analysis of The Argumentation Framework with Subargument Relation ", supervised by Prof. Beishui Liao. Her research interests focus on logic and formal argumentation, it also covers their applications on risk analysis and regulatory compliance of distributed ledger technologies for transaction and management of securities.
Page updated
Google Sites
Report abuse