April 14, 2022 - Submissions Due April 30, 2022 - Author Notification
- May 31, 2022 - Camera-ready Due
- June 17, 2022 - Workshop Date
Data-driven AI is the dominating trend in AI at this time. From a planning and scheduling perspective – and for sequential decision making in general – this is manifested in two major kinds of technical artifacts that are rapidly gaining importance. First, planning models learned from data, or partly learned from data (such as e.g. a weather forecast in a model of flight actions). Second, action-decision components learned from data, in particular, action policies or planning-control knowledge for making decisions in dynamic environments (such as e.g. manufacturing processes under resource-availability and job-length fluctuations). Given the nature of such data-driven artifacts, reliability is a key concern, prominently including safety, robustness, and fairness in various forms, but possibly other concerns as well. Arguably, this is indeed one of the grand challenges in AI for the foreseeable future.
Given this, the workshop welcomes contributions to any topic that roughly falls into the following problem space:
- Data-driven artifacts: Reliability of learned planning and scheduling models (e.g. action models, transition probabilities, environment prediction, etc.); learned action-decisions (e.g. action policies, components thereof, previous plans, etc.); combinations of both.
- Objectives: Reliability in whatever form, including risk, safety, robustness, fairness, error bounds, etc.; alongside possibly other concerns such as scalability and data efficiency, system design/engineering principles and challenges, and the interactions of these with reliability.
- Methodologies: Planning and scheduling algorithms in the presence of learned artifacts as per 1.; analyzing such artifacts (reasoning, verification, testing, etc.); making such analyses amenable to human users (visualization, interaction); potentially others as relevant to the objectives as per 2.
We call for two kinds of submissions:
- Technical papers, of length up to 8 pages plus references. The workshop is meant to be an open and inclusive forum, and we encourage papers that report on work in progress.
- Position papers, of length up to 4 pages plus references. Given that reliability of data-driven planning and scheduling is rather new at ICAPS, we encourage authors to submit positions on what they believe are important challenges, questions to be considered, approaches that may be promising. We will include any position relevant to discussing the workshop topic. We expect to group position paper presentations into a dedicated session, followed by a panel discussion.
Please do not submit papers that are already accepted for the main conference. All other submissions, e.g. papers under review for IJCAI'22, are welcome. Authors submitting papers rejected from the main conference, please ensure you do your utmost to address the comments given by ICAPS reviewers. Also, it is your responsibility to ensure that other venues your work is submitted to allow for papers to be already published in "informal" ways (e.g. on proceedings or websites without associated ISSN/ISBN).
Every submission will be reviewed by members of the program committee according to the usual criteria such as relevance to the workshop, significance of the contribution, and technical quality.
At least one author of each accepted paper must attend the workshop in order to present the paper. The workshop format (fully virtual or hybrid) will be the same as the format of the main conference. Authors must register for the ICAPS conference in order to attend the workshop.
- Submission Deadline: 14 April 2022, AoE
- Author Notification: 30 April 2022
- Camera-Ready Deadline: 31 May 2022, AoE
Differential Assessment of Black-Box AI Agents
Rashmeet Kaur Nayyar, Pulkit Verma and Siddharth Srivastava
NN Policy Verification via Predicate Abstraction: CEGAR
Marcel Vinzent and Jörg Hoffmann
Planning with Dynamically Estimated Action Costs
Eyal Weiss and Gal Kaminka
Metamorphic Relations via Relaxations: An Approach to Obtain Oracles for Action-Policy Testing
Hasan Ferit Enişer, Timo Gros, Valentin Wüstholz, Jörg Hoffmann and Maria Christakis
An AI Safety Threat from Learned Planning Models
Toryn Q. Klassen, Sheila A. McIlraith and Christian Muise
Position Paper: Online Modeling for Offline Planning
Eyal Weiss and Gal Kaminka
Neural Network Heuristic Functions: Taking Confidence into Account
Daniel Heller, Patrick Ferber, Julian Bitterwolf, Matthias Hein and Jörg Hoffmann
Collaborative Multi-Agent Planning with Black-Box Agents by Learning Action Models
Argaman Mordoch, Daniel Portnoy, Brendan Juba and Roni Stern
Scaling up ML-based Black-box Planning with Partial STRIPS Models
Matias Greco, Álvaro Torralba, Jorge A. Baier and Hector Palacios
Schedule — Friday, June 17, 2022
|Canberra||Paris||New York||Los Angeles|
|Opening Remarks||11:55 pm 12:00 am||3:55 pm 4:00 pm||9:55 am 10:00 am||6:55 am 7:00 am|
|Session 1: Invited Talk: Florent Teichteil-KönigsbuchChair: Jörg Hoffmann||12:00 am 12:50 am||4:00 pm 4:50 pm||10:00 am 10:50 am||7:00 am 7:50 am|
|Session 2: Quality assurance of learned action policiesChair: Sylvie Thiebaux||12:50 am 1:30 am||4:50 pm 5:30 pm||10:50 am 11:30 am||7:50 am 8:30 am|
|Session 3: Improving learned search informationChair: Alan Fern||1:30 am 2:10 am||5:30 pm 6:10 pm||11:30 am 12:10 pm||8:30 am 9:10 am|
|Break||2:10 am 2:40 am||6:10 pm 6:40 pm||12:10 pm 12:40 pm||9:10 am 9:40 am|
|Session 4: Learned planning modelsChair: Jörg Hoffmann||2:40 am 3:20 am||6:40 pm 7:20 pm||12:40 pm 1:20 pm||9:40 am 10:20 am|
|Session 5: Position papersChair: Jörg Hoffmann||3:20 am 4:10 am||7:20 pm 8:10 pm||1:20 pm 2:10 pm||10:20 am 11:10 am|
|Session 6: Panel discussion "RDDPS Quo Vadis?"Chair: Jörg Hoffmann||4:10 am 5:00 am||8:10 pm 9:00 pm||2:10 pm 3:00 pm||11:10 am 12:00 pm|
Attending RDDPSAs ICAPS22 has decided to go fully virtual, RDDPS will be completely virtual as well. Further details will be announced in time.
- Jörg Hoffmann, Saarland University, Germany
- Sara Bernardini, Royal Holloway University of London, UK
- Michele Lombardi, DISI, University of Bologna, Italy
- Alan Fern, Oregon State University, USA
- Hector Palacios, ServiceNow Research, Canada
- Scott Sanner, University of Toronto, Canada
- Sylvie Thiebaux, The Australian National University, Australia
- Marcel Steinmetz, Saarland University, Germany