Electrical & Computer Engin 697CM - ST-InspirationalCognitiveMach

Is understanding the brain's design essential for inspiring artificial cognitive machines? Shall we focus on emulating its neural circuitry bottom-up, hoping it will do the magic at large scale, or emulating its functions and integration, top-down? Is consciousness necessary for artificial general cognition? Is the technology of the brain just one implementation? To discuss, we will start with exploring the machinery of the brain based on published books and papers - theories for long-term memory, models of consciousness and active perception, language encoding with universal grammar, quantitative reasoning based on number sense, conceptual integration, and cognitive processes for morals, from intuition to judgement and reason. Inspired from this, we will look at how cognition could work artificially, what architectural directions exist or may be possible. We touch upon the relevance of black-box machine learning, white-box AI-like circuit architectures with Bayesian Networks and statistical inference, neuron-like function-equivalent adaptable circuitry, technologies for trillion+ devices for detailed brain circuit emulation, and opportunities with quantum computing. The course explores many open questions and is at the intersection of cognitive science, physical sciences, and computing with an eye towards inspiring a broad view on cognitive engineering. It is structured with a mix of lectures, research papers and book discussions, and presentations by students.

Open to graduate Computer Science and Electrical Engineering students only. Familiarity with concepts in AI/ML, Computer Architecture, and some VLSI. Interest or familiarity in cognitive and neuroscience.