HereBot: An Artificial Curiosity Engine for One-Shot Language Learning in a Non-Transformer Cognitive Architecture
Description
Artificial intelligence today is mainly trained on transformer architecture and later fine-tuned along
with data segregation requiring enormous training data and machine intensive processes . Unlike a
traditional transformer, the research of Herebot started with a simple question: How do cognitive
beings actually learn some concepts in scoped or classified context? And can that same mecha-
nism be applied in order to learn within machine learning concepts rather than using a transformer
concept? The main difference where the transformer and Herebot differ is, transformer does encod-
ing and embedding, tokenization etc. processes which includes heavy computational processing.
Whereas an alternative architecture can hardcode documented learning traits of cognitive beings
and use them instead of transformer-like training to make or justify the learning process. The
research due to its complexity currently is focused on textual data and theories on cognition and
memory. As biologically and psychologically the cognitive being’s mind and its working isn’t
fully documented yet, but even though a major portion is theoretics (theory papers as cited) exists,
computer algorithms from these were derived for Herebot.
Files
paper.pdf.pdf
Additional details
Software
- Repository URL
- https://github.com/hrik3/herebot
- Programming language
- Python , Python console
- Development Status
- Active
References
- Aubret, A., Matignon, L., & Hassas, S. (2023). An information-theoretic perspective on intrinsic motivation in reinforcement learning: A survey. Entropy, 25(2), 199.
- Itti, L., & Baldi, P. (2009). Bayesian surprise explains human attention. Neural Computation, 21(11), 2605–2631
- McClelland, J. L., McNaughton, B. L., & O'Reilly, R. C. (1995). Why there are complementary learning systems in the hippocampus and neocortex. Psychological Review, 102(3), 419– 457
- Nagashima, K., Morita, J., & Takeuchi, Y. (2024). Intrinsic motivation in cognitive architecture: intellectual curiosity originated from pattern discovery. Frontiers in Artificial Intelligence, 7, 1397860.
- Oudeyer, P.-Y., & Kaplan, F. (2007). What is intrinsic motivation? A typology of computational approaches. Frontiers in Neurorobotics, 1, 6.
- Rahimi, A., & Recht, B. (2007). Random features for large-scale kernel machines. Advances in Neural Information Processing Systems, 20
- Schmidhuber, J. (1991). A possibility for implementing curiosity and boredom in model-building neural controllers. In Proceedings of ICANN (pp. 145–151)
- Schmidhuber, J. (2006). Developmental robotics, optimal artificial curiosity, creativity, music, and the fine arts. Cognitive Systems Research, 7(4), 251–268
- Teh, Y. W., Jordan, M. I., Beal, M. J., & Blei, D. M. (2006). Hierarchical Dirichlet processes. Journal of the American Statistical Association, 101(476), 1566–1581.