AAAI-21 Focus Areas
Beyond the usual keywords, the AAAI-21 main track will highlight three focus areas. The Conference Committee especially encourages submissions in these focus areas, and will highlight these at the conference in separate sessions.
Focus Area on Neuro-Symbolic AI
A grand challenge for artificial intelligence is to develop methods that can combine learning, reasoning, optimization, and constraints within a single unified framework. In recent years, the community has developed a number of exciting new approaches that address aspects of this challenge, often combining aspects of differentiable learning with more symbolic, discrete, or optimization-based components. Often this is phrased as approaches that combine the traditional “System 1” (low-level perception, or sub-symbolic) and “System 2” (high-level reasoning, or symbolic) aspects of artificial intelligence.
This focus area call encourages submissions that seek to address this challenge, interpreted broadly as methods that combine traditional learning-based approaches with more structured reasoning, optimization (both continuous and combinatorial), logical, or other symbolic approaches.
Example topics include, but are not limited to,
- Representation learning over trees, graphs, geometric objects, first-order logic formulas, and other symbolic objects
- Using intermediate symbolic representations within a neural network
- Injecting world knowledge into deep neural networks
- Neural networks with prior symbolic knowledge, such as constraints, symbolic features, etc.
- Incorporation of traditional solvers as layers in differentiable architectures
- (Deep) probabilistic (logic) programming
- Hybrid methods, combining traditional symbolic algorithms along with neural models
- Investigations into the tractability of joint neural and symbolic architectures
The submissions for this focus area will follow the regular AAAI technical paper submission process. Submissions to this area will be reviewed by a program committee of experts in this area.
Accepted papers will be presented at the main conference in sessions designated to the topic. The sessions may also include invited talks and panels on the topic.
Focus Area Chairs
Kristian Kersting (TU Darmstadt, Germany)
Zico Kolter (Carnegie Mellon University, USA)
Focus Area on AI Responses to the Covid-19 Pandemic
AAAI 2021 will feature a separate session in the main conference on AI Responses to the COVID-19 Pandemic. The intent is to recognize and promote high-quality research work that is directly related to fighting the COVID-19 pandemic, whether addressing its medical, social, or economic consequences.
The focus area session is motivated by current, pressing needs, including but not limited to:
- a lack of effective, reliable, and low-cost means for COVID-19 screening;
- privacy concerns arising from contact tracing;
- understanding the causes of patient mortality and and identifying prognosis models among different patient populations;
- comprehensive in-hospital patient profiling via multi-sensory data to assist optimized treatment planning and outcomes.
The goal of this focus area session is to highlight these technical challenges and opportunities and to showcase the strengths of artificial intelligence when applied to the fight against COVID-19. Topics include, but are not limited to,
- efforts to further scientific understanding of the virus and its impact on society;
- models of the spread of the pandemic;
- approaches leveraging AI to help contain the virus;
- imaging and non-imaging methods for improving the screening, diagnosis, and treatment of COVID-19 patients;
- mitigations for economic, social or technical challenges arising from the pandemic;
- distributed, privacy-preserving AI computing methods.
Orientation toward real solutions
The focus area particularly welcomes solutions that are oriented toward real needs. We thus particularly seek papers that present:
- solutions to address core needs related to COVID-19;
- solutions that have sustainable life cycles;
- solutions that can be implemented at low cost;
- solutions that can be utilized as clinical decision-making tools for physicians and clinicians;
- new AI frameworks or tools that can be useful to epidemiologists;
- solutions that permit quantitative precision imaging biomarkers in longitudinal patient studies;
- solutions that can be used to monitor, provide surveillance on patient disease progression and suggest intervening measures to optimize patient prognosis;
- solutions that are able to understand, predict and model the COVID-19 pandemic, including potential future outbreaks;
- solutions that can be merged with current systems or are otherwise compatible with existing technologies;
- open source algorithms, models, and large-scale datasets that accelerate AI research for fighting COVID-19.
Submissions to this focus area will follow the regular AAAI technical paper submission procedure through the same CMT system. To be considered for the focus area, the submission should select the designated focus area keyword (‘AI Responses to the COVID-19 Pandemic’’). Papers submitted to this focus area will be reviewed according to the same standards and procedures as other AAAI technical submissions, but overseen by the dedicated focus area chairs.
Focus Area Chairs
S. Kevin Zhou (Chinese Academy of Sciences, China)
Le Lu (PAII Inc. Bethesda Lab, USA)
B. Aditya Prakash (Georgia Tech, USA)
Focus Area on AI for Conference Organization and Delivery
AAAI 2021 will feature a focus area on AI for Conference Organization and Delivery. This focus area recognizes the urgent need for developing AI-based technologies that aid in the organization and delivery of large conferences, such as AAAI itself, and welcomes high-quality submissions that address the various challenges involved.
The focus area is motivated by the following observations:
- Large conferences are getting larger every year and becoming increasingly difficult to manage using traditional approaches
- AI-based technologies are already being used to aid various steps of the process, such as assigning reviewers to submissions, but are relatively understudied
- Due to the COVID-19 pandemic, many conferences have suddenly moved to virtual delivery, which has given rise to unique challenges
- Because virtual delivery has its own advantages (e.g. potential for broader participation), it is likely to survive in some form beyond the COVID-19 pandemic
This focus area invites high-quality submissions that study various ways in which AI-based technologies can aid in the organization and delivery of conferences, including but not limited to:
- Aiding authors: discovering related work, generating literature review, formatting and format checking
- Aiding the review process: bidding recommendations, assigning reviewers to papers, assessing paper and review quality, facilitating reviewer discussion, incentivizing and facilitating high-quality reviews
- Aiding conference organization: partitioning papers into sessions, scheduling sessions while avoiding conflicts
- Aiding attendees: recommending talks, summarizing presentations, tools for meeting people and having productive conversations
- Aiding presenters: assessing audience attention and understanding, summarizing audience feedback, moderating audience questions
This focus area particularly welcomes submissions that have one or more of the following qualities:
- Identifies a novel problem domain
- Proposes a novel approach
- Offers sound theoretical justification
- Conducts robust empirical evaluation
- Shows promise for quick and widespread adoption in practice
Submissions to this focus area will follow the regular AAAI technical paper submission procedure through the same CMT system. To be considered for the focus area, submissions must select the designated focus area keyword (“AI for Conference Organization and Delivery’). Papers submitted to this focus area will be reviewed according to the same standards and procedures as other AAAI technical submissions, but overseen by the dedicated focus area chairs.
Questions and Suggestions
For concerns, questions, and suggestions regarding this focus area, write to the focus area chairs:
David R. Karger (Massachusetts Institute of Technology, USA)
Nisarg Shah (University of Toronto, Canada)