Neuro-symbolic AI
We see Neuro-symbolic AI as a pathway to achieve artificial general intelligence. By augmenting and combining the strengths of statistical AI, like machine learning, with the capabilities of human-like symbolic knowledge and reasoning, we're aiming to create a revolution in AI, rather than an evolution.
Our work
Disentangling visual attributes with neuro-vector-symbolic architectures, in-memory computing, and device noise
Technical noteAbu Sebastian, Abbas Rahimi, and Geethan KarunaratneThis AI could likely beat you at an IQ test
ResearchAbbas Rahimi and Michael HerscheHow IBM Research is accelerating discoveries in the fight against COVID-19
ResearchMike Murphy10 minute readAI, you have a lot of explaining to do
ReleaseDinesh Garg, Parag Singla, Dinesh Khandelwal, Shourya Aggarwal, Divyanshu Mandowara, and Vishwajeet Agrawal5 minute readIBM, MIT and Harvard release “Common Sense AI” dataset at ICML 2021
ReleaseDan Gutfreund, Abhishek Bhandwaldar, and Chuang Gan6 minute readMimicking the brain: Deep learning meets vector-symbolic AI
ResearchAbu Sebastian and Abbas Rahimi4 minute read- See more of our work on Neuro-symbolic AI
Projects
Accelerator technologies
Publications
- Michael Katz
- Junkyu Lee
- et al.
- 2023
- SoCS 2023
- Zijun Cui
- Chenyi Kuang
- et al.
- 2023
- CVPR 2023
- Georgios Kollias
- Vasileios Kalantzis
- et al.
- 2023
- ICASSP 2023
- Lee Martie
- Jessie Rosenberg
- et al.
- 2023
- ICSE 2023
- Yi Zhou
- Parikshit Ram
- et al.
- 2023
- ICLR 2023
- Alex Gu
- Songtao Lu
- et al.
- 2023
- ICLR 2023
Neuro-symbolic AI research at the MIT-IBM Watson AI Lab
Read more about our work in neuro-symbolic AI from the MIT-IBM Watson AI Lab. Our researchers are working to usher in a new era of AI where machines can learn more like the way humans do, by connecting words with images and mastering abstract concepts.
Tools + code
Transition-based AMR Parser
Transition-based parser for Abstract Meaning Representation (AMR) in Pytorch.
View project →IBM Hyperlinked Knowledge Graph
A set of libraries to provide an abstraction layer to access the endpoints of databases that store IBM Hyperlinked Knowledge Graphs.
View project →Forbid-Iterative Planner
Forbid-Iterative Planner is an automated PDDL-based planner that includes planners for top-k, top-quality, and diverse computational tasks.
View project →ITOPS: an ontology for IT Operations
Ontology source files for the IT Operations Ontology derived from Wikidata, DBPedia and Wikipedia.
View project →Classical planner FD-Novelty-PO
Code for IJCAI 2021 paper “The Fewer the Merrier: Pruning Preferred Operators with Novelty”
View project →