Coursera Course Computational Neuroscience, lecture 7.3 Sparse Coding and Predictive Coding
How to use:
>octave
>cd <dir>
>cd regression-analysis
>network
>
>cd ..
>cd system-dynamics
>network
Number of differential equations: ( N * S + N * M) where: N = number of neurons, M = number of inputs, S = number of samples
The results clearly show convergence to sparseness.
Issues:
- The implemented sparseness constraint is different from that in the lecture.
- It is not a system dynamics model, it does not model the physics of neurons in a network
Typical results:
Number of differential equations: ( N + N * M) where: N = number of neurons, M = number of inputs
Issues:
- It is not clear what this algorithm is doing
- The learning capability of this algorithm is not clear
- The effect of the sparseness constraint is not clear
- The neurons are modelled as one single system with multiple outputs, the neurons seem to share each others hidden state
Typical results:
