VLAR 2023 - Vision-and-Language Algorithmic Reasoning

About VLAR 2023

The focus of this workshop is to bring together researchers in multimodal reasoning and cognitive models of intelligence, towards positioning the current research progress in AI within the overarching goal of achieving machine intelligence. An important focus of this workshop is to bring to the forefront problems in perception, language modeling, and cognition that are often overlooked in state-of-the-art research and that are important for making true progress in AI. One specific problem that motivated our workshop is the question of how well current deep models learn broad yet simple skills and how well do they generalize their learned models to solve problems that are not part of their learning set; such skills even children learn and use effortlessly. We attempt to look into this aspect of intelligence in the CVPR 2023 paper titled: Are Deep Neural Networks SMARTer than Second Graders? In this workshop, we plan to bring together outstanding faculty/researchers working at the intersections of vision, language, and cognition to provide their opinions on the recent breakthroughs, as well as showcase their cutting edge research on the above topics that could inspire the audience to search for the missing pieces in our quest for artificial intelligence.

Where

Paris Convention Centre (Details TBD)

When

October 3, 2023 (exact time TBD)

Keynote Speakers

Anima Anandkumar

NVIDIA & Caltech

François Chollet

Google

Jitendra Malik

Meta & UC Berkeley

Elizabeth Spelke

Harvard University

Jiajun Wu

Stanford University

VLAR 2023 Schedule

[in CET (GMT+1)]

[More info about the schedule will be updated here]

Submission

[call for contributions PDF]

Paper Track: Submission Instructions

We welcome paper submissions up to 4 pages (excluding references or supplementary materials). Please submit at the
VLAR 2023 @ ICCV 2023 CMT website.

The paper submissions must be in pdf format and use the official ICCV 2023 templates. All submissions must be anonymous and conform to ICCV standards for double-blind review. The accepted papers will be included in the ICCV 2023 proceedings. The accepted papers will be presented as either an oral, spotlight, or poster presentation. At least one author of each accepted submission must present the paper at the workshop. The presentation of the accepted papers at VLAR 2023 will follow the same policy as that for the accepted papers of ICCV 2023.

Submission deadline: July 20, 2023 (11:59PM EDT).
Notification to authors: August 1, 2023.
Camera ready deadline: August 9, 2023 (11:59PM EDT).

We invite the submission of original and high-quality research papers in the topics related to vision-and-language algorithmic reasoning. Accepted work will be presented as either an oral, spotlight, or poster presentation.

Paper Track: Topics

The topics for VLAR 2023 include, but are not limited to:

Large language models, vision, and cognition including children’s cognition.
Foundation models of intelligence, including vision, language, and other modalities.
Artificial general intelligence / general-purpose problem solving architectures.
Neural architectures for solving vision & language or language-based IQ puzzles.
Embodiment and AI.
Large language models, neuroscience, and vision.
Functional and algorithmic / procedural learning in vision.
Abstract visual-language reasoning, e.g., using sketches, diagrams, etc.
Perceptual reasoning and decision making.
Multimodal cognition and learning.
New vision-and-language abstract reasoning tasks and datasets.
Vision-and-language applications.

SMART-101 Challenge Track: Participation Instructions

As part of VLAR 2023, we are hosting a challenge based on the Simple Multimodal Algorithmic Reasoning Task – SMART-101 – dataset. The accompanying CVPR 2023 paper is: “Are Deep Neural Networks SMARTer than Second Graders?” The challenge will be hosted on Eval AI. The challenge participants are required to make arXiv submissions detailing their approach. These are only used to judge the competition, and will not be reviewed and will not be part of workshop proceedings. The winners of the challenge are determined both by performance on the leaderboard over a private test set as well as the novelty of the proposed method (as detailed in the arXiv submission). The details will be made available on the challenge website. The prizes will be awarded to the winners on the day of the workshop. The key dates of the challenge are as follows:

Challenge open: June 15, 2023.
Submission deadline and arXiv paper deadline to be considered for awards: September 1, 2023 (11:59PM EDT).
Public winner announcement: October 3, 2023 (11:59PM EDT).