SESSION 8: ARTIFICIAL NEURAL NETWORKS

 

 

Exemplars from class

Letter Definitions  (14-feature vectors for letter reps from Rumelhart & McClelland)

ReuterÕs Articles (corporate takeovers; see Thurson JoachimÕs site for description)

Fingerprint vectors  (and README from NIST)

Number vectors (from handwritten numbers. Binary pixels from 64 x 40 pixel fieldÉeach 64-pixel row is concatenated in a single line per exemplar; first character is identity; 36k exemplars, and all have been sheared, cropped, and centered)

Elman artificial language corpus

Wikipedia corpus sample (sports, literature, biology samples)

 

 

Slides

Slides Part 1 (Intro to ANNs)

Slides Part 2 (Backprop)

Backprop Pseudocode

 

Code

README

Number_Generators.f95

Perceptron.f95

Exemplars.f95

Prototype.f95

Backprop.f95

 

 

BACK TO SCHEDULE