Latest paper is here: https://arxiv.org/abs/1805.09001
code for figures and experiments
Figure 3: typical lambda function and their f.p.
conn_09_typical_funcs_convergence.py with different plasticity functions PF32, PF12, PF30, PF15.
Figure 4: one-to-one mapping from stimulus to strength
conn_08_s_x_relation_pf_30.py
conn_08_s_x_relation_pf_32.py
Figure 5: different stimulus
conn_09_typical_funcs_convergence.py with plasticity functions PF12.
conn_10_s_x_relation_sin_pf.py to plot theta function.
Figure 6: Theta function of discontinuous lambda
conn_14_s_x_relation_discont_pf.py
Figure 7: lambda of choice
conn_12_pf_of_year.py constructs linear and threshhold-like thelta functions.
Figure 9: neural network to f.p. on different lambda
nn_02_different_e_to_fp.py with different plasticity functions PF32, PF12, PF30, PF15.
Figure 12: meshed-NN classifier for digit 6
nn_meshed_dist_6_digit.py
Figure 13: meshed-NN classifier for all digits
nn_meshed_dist_digit.py
Figure 15: Optimization of growable classifier
nn_growable_6_digit_80_81.py
meshed-NN
training: nn_meshed_train_batch.sh
testing: nn_meshed_test_batch.sh
per-digit testing: nn_meshed_test_number_batch.sh
growable-NN
training: nn_growable_train_batch.sh
testing: nn_growable_test_batch.sh