Xiaoxi Wang 2016-01-18

Last week:

Neural Turing Machine for arithmetic calculation (N number addition)

coding for NTM mini-batch training

This week:

review papers on neural arithmetic calculation

debug NTM mini-batch training

Some output from NTM:

Number 1: [ 0. 0. 1. 1. 1. 0. 0. 0. 0.] (28) Number 2: [ 1. 0. 0. 0. 0. 0. 0. 0. 0.] (1) Output: [ 9.99984384e-01 1.68709391e-06 9.99961972e-01 9.99988675e-01 9.99942899e-01 4.12209111e-08 6.87969441e-06 1.36186891e-05 4.81172719e-07] Result: [ 1. 0. 1. 1. 1. 0. 0. 0. 0.] (29)

Number 1: [ 0. 1. 1. 1. 0. 1. 0. 0. 0.] (46) Number 2: [ 1. 1. 0. 1. 0. 0. 0. 0. 0.] (11) Output: [ 9.99987006e-01 1.28979622e-07 3.68938090e-05 9.99997497e-01 9.99983549e-01 9.99990582e-01 7.21790057e-05 4.65510175e-06 3.80014421e-06] Result: [ 1. 0. 0. 1. 1. 1. 0. 0. 0.] (57)