PhD Topics Chris Huyck is interested in Supervising
Computational Cognitive Neuroscience PhD Topics
- Associative Memory with Cell Assemblies: we've worked with Cell
Assemblies for decades now. While we can build associative memories,
we are interested in memories that are learned. Moreover, the
short-term memories should persist for times similar to human
short-term memories. This could be done as part of the Human
Brain Project mechanism with PyNN and SpiNNaker or Nest. It could
also be done with my java FLIF simulator CANT.
- Improved CABot3 Vision: our existing virtual agent works in
the java simulator and on the SpiNNaker HBP platform. Vision
is rudimentary with line dectectors and object detectors. There
is wide scope to improve this mechanism. This could use learning
to learn new visual objects; implement depth perception, for example
with binocular vision; learn early visual features; or a wide range
of other neuro-biologically plausible improvements.
- Cognitive Mapping for CABot3: our CABot3 agent has a very simple
cognitive mapping mechanism. A radically improved mechanism could
be developed. This could take advantage of grid or place cells.
Memory could be transient.
- I'm open to discussion of other Computational Cognitive Neuroscience
PhD Topics. However, I all but insist on some neural implementation.
- I'm interested in language processing with neurons. In particular,
cascaded finite state automata are flexible fast mechanisms that could
be easily implemented on neuromorphic hardware.
- I'm interested in neural deep belief nets. We have done some work
with multi area learning with biologically plausible mechanisms.
This could easily be expanded.
Natural Language Processing Topics
- I'm interested in chatbots. A domain specific chatbot implemented
in existing technologies could address a range of interesting topics.
- I'm interested in language processing with neurons. See above.
This page was developed by
Page last modified