I explore the frontier between symbolic reasoning and deep learning, working toward more general and interpretable forms of artificial intelligence. My current focus is on working with non-axiomatic reasoning systems (NARS) — bridging logic, learning, and resource-bounded cognition.
- Reasoning Systems: Built experiments like NARCA — a NARS-driven game agent showcased at AGI-22 — and NARpyn, a Python API that makes NARS For Applications easier to use.
- Julia for AI: I want to see how far can I go with Julia-based implementations, like NACE.jl Julia provides easy performance gains over Python and other useful features.
I am interested in neuro-symbolic AI — algorithms alternative to the typical sub-symbolic approach or, preferably, existing somewhere between the symbolic and sub-symbolic divide; to go beyond the limitations inherent to Artificial Neural Networks. My current interests include:
- experimenting with probabilistic/symbolic AI systems,
- integrating neural models with reasoning agents, and
- contributing open, reproducible tools for AGI research.