> The HTM is Jeff's over-arching theory of how the neocortex works. The CLA is > a very detailed model (suitable for implementation in software right now) > which describes how Layer 3 of a single region recognises spatial patterns > and learns sequences of those patterns. The CLA (and NuPIC) encompasses three > of the six principles of HTM, namely Sparse Distributed Representations, > Online Learning, and Sequence Memory. Hierarchy, Attention and Motor Function > are TBD right now! > Because of the lack of hierarchy in NuPIC, it cannot compete with other "some > neuroscience" approaches such as Deep Belief Nets for applications such as > yours. However, these other approaches lack the temporal sequence learning > which is one of the core attributes of NuPIC.
Thank you for the explanation! This is exactly why I'm writing a small library that enables hierarchy (c++ at the moment for speed). Basically I have created some code that is relatively fast and mimics CLA in some respect (I don't have dutyCycle watcher for boosting at the moment, and it's currently does only spacial pooling). The reason I do this to begin with, is for easier translation to GPU, and a little glitch of mine - for me it's much easier to write something myself then understand how a huge library works. It is only when I fail to achieve the result I turn back to the library, but by then I already understand everything. > For categorisation of static images, I would look at Geoff Hinton's work (he > has a great, and free, course on Udemy). The system he describes looks very > much like what you're doing. I found one on Coursera https://class.coursera.org/neuralnets-2012-001/lecture/index Is this the one? Looks like it has invaluable information there! > Time-based learning in NuPIC would involve providing a region with a varying > input and have it learn the invariance. This would not be done by giving it a > set of real-world, unrelated images, but perhaps by using 3D rendering > software to feed it a realistic feed of successive, semantically consistent > frames, just like what we get when we perceive the world. I still think we'll > have to wait until we have a hierarchy of CLA's before we get general vision > happening here. 3D rendering is a great idea! _______________________________________________ nupic mailing list [email protected] http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
