Multi-Objective and Multi-Task Optimization for Robust AI

Overview

We invite researchers, practitioners, and students to participate in the AAAI-27 Workshop on Multi-Objective and Multi-Task Optimization for Robust AI, a one-day workshop at the Forty-First AAAI Conference on Artificial Intelligence in Montréal, Canada. The workshop will take place on February 22 or 23, 2027; the exact date will be assigned by AAAI.

Trustworthy AI is inherently multi-dimensional. Modern AI systems must remain accurate and useful while also being robust to adversarial perturbations and distribution shift, safe, fair, privacy-preserving, transparent, reliable, and efficient. These requirements often interact or conflict, and a single scalar score can obscure the trade-offs that matter for training, evaluation, and deployment.

The workshop connects multi-objective optimization (MOO), evolutionary computation, multitask learning, and robust and trustworthy AI. We will explore how Pareto-aware optimization, hard constraints, preference elicitation, and knowledge transfer across tasks can address competing trustworthiness requirements in foundation models, reinforcement learning, and autonomous systems.

With a central focus on multi-objective and multi-task optimization, we welcome theory, algorithms, applications, and evaluations, including related work that motivates new optimization formulations. Contributions from early-career researchers and across disciplines are particularly encouraged.

Important Dates

• Paper Submission Deadline: November 20, 2026 (AoE)
• Notification of Acceptance: December 2, 2026 (AoE)
• Camera-Ready Deadline: Jan 2, 2027 (AoE)
• Workshop Date: Feb 22–23, 2027 (exact day TBC)

Speakers

More invited speakers will be announced soon...

Program

Time (EST)	Topic
TBD	Opening Remarks (Organizers)
TBD	Invited Talk 1 (Speaker TBD)
TBD	Invited Talk 2 (Speaker TBD)
TBD	Invited Talk 3 (Speaker TBD)
TBD	Coffee Break
TBD	Invited Talk 4 (Speaker TBD)
TBD	Invited Talk 5 (Speaker TBD)
TBD	Invited Talk 6 (Speaker TBD)
TBD	Invited Talk 7 (Speaker TBD)
TBD	Lunch Break
TBD	Oral Presentations (Selected from submissions)
TBD	Poster Session (Selected from submissions)
TBD	Panel Discussion & Closing Remarks (All Speakers and Organizers)

Call for Papers

We invite submissions of short papers (4 pages) and full papers (8 pages), excluding references and appendices. All submissions should follow the AAAI format.

Topics of interest include but are not limited to:

• Multi-objective optimization for trustworthy AI
• Pareto-aware training and evaluation
• Evolutionary computation for robust AI
• Multi-task learning and optimization
• Adversarial robustness via multi-objective methods
• Fairness-accuracy trade-offs
• Privacy-utility optimization
• Constraint-handling in AI safety
• Foundation model alignment through optimization
• Benchmarks for multi-objective AI evaluation

Submission site: OpenReview - AAAI 2027 Workshop MultiRobust

The workshop will be held in a hybrid format, allowing both in-person and virtual participation. At least one author of each accepted paper must register for the workshop and present their work. Virtual/Remote Attendance will be available to everyone who has registered for the event.

Organizers

Contact

Workshop inquiries: Junhao Dong, Nanyang Technological University.
Email: junhao003@ntu.edu.sg.