A naukri.com initiative
Hackernoon
6d
70
Image Credit: Hackernoon
AI That Remembers
- Researchers at Carnegie Mellon University and Princeton University have created an AI system that can recall previous experiences from a larger dataset and utilize it for future tasks.
- The team introduced a selective state-space model (SSM) that uses a selection mechanism to recall only relevant information, helping to reduce computational power and memory requirements.
- The system demonstrated success in handling synthetic tasks, audio modeling and generation, DNA modeling and language modeling with larger memory compared to previous approaches.
- The AI studied in the work is known as a state-space neural network (SSNN).
- According to the team, the AI can perform much faster when it is not bogged down trying to remember irrelevant information.
- Previous state-space models typically used the LTI (linear time-invariant) and non-selective structured algorithm while the proposed model is better at handling large datasets.
- The researchers' findings offer a new way of storing data in an AI and may lead to future AIs capable of autonomously remembering past experiences.
- The AI could have implications in healthcare, with machine learning systems capable of diagnosing and handling illnesses by analyzing medical data.
- The team is now exploring how the AI could be improved and used in wider applications.
- The study work is available under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International license.
Read Full Article
4 Likes
For uninterrupted reading, download the app