Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

Enabling Lifelong Learning in AI with Biological Neural Networks Based on Short-Term, Working, and Long-Term Memory

Hanav Modasiya¹; ¹Santa Clara High School

Presenter: Hanav Modasiya

Achieving Lifelong Learning, the ability of a learning system to continuously acquire and adapt to changing data over time, in Artificial Intelligence (AI) is an integral step towards achieving Artificial General Intelligence (AGI), a hypothetical version of AI that can change our world forever. Currently, the vast majority of models are incapable of exhibiting Lifelong Learning. This research theorizes the first Neural Network architecture inspired by the Three-stage Memory Model--a theory on our brain's memory. By developing a complementary Neural Network learning system comprising the Cerebral Cortex, Prefrontal Cortex, and Hippocampus, mimicking Long-term, Working, and Short-term memory, respectively, this research achieves Lifelong Learning on a simulated computer vision task and develops the first Working memory-inspired model. It also demonstrates the feasibility of using the Three-stage Memory Model for achieving human-like cognition in AI. Therefore, this research reveals a pathway for future research in achieving AGI: the Three-stage Memory Model.

Topic Area: Memory, Spatial Cognition & Skill Learning

Extended Abstract: Full Text PDF