Collaboration diagram for Training a plain unmodulated network:

Functions
	BOOST_AUTO_TEST_CASE (trainparams)
	Test training. This just checks that the network trains. More...

	BOOST_AUTO_TEST_CASE (trainparams2)
	another test without cross-validation which attempts to emulate the Angort test.ang program. More...

	BOOST_AUTO_TEST_CASE (addition)
	[addition] More...

	BOOST_AUTO_TEST_CASE (additionmod)
	[addition] More...

	BOOST_AUTO_TEST_CASE (trainmnist)
	[additionmod] More...

Detailed Description

Function Documentation

BOOST_AUTO_TEST_CASE ( trainparams )

Test training. This just checks that the network trains.

Definition at line 25 of file testTrainBasic.cpp.

BOOST_AUTO_TEST_CASE ( trainparams2 )

another test without cross-validation which attempts to emulate the Angort test.ang program.

Definition at line 72 of file testTrainBasic.cpp.

BOOST_AUTO_TEST_CASE ( addition )

[addition]

Construct an addition model from scratch and try to learn it with backprop

Definition at line 118 of file testTrainBasic.cpp.

BOOST_AUTO_TEST_CASE ( additionmod )

[addition]

[additionmod] Construct an addition/addition+scaling model from scratch and try to learn it with UESMANN. The h=0 is $y=a+b$ , the h=1 function is $y=(a+b)*0.3$ .

Definition at line 201 of file testTrainBasic.cpp.

BOOST_AUTO_TEST_CASE ( trainmnist )

[additionmod]

[trainmnist] Train for MNIST handwriting recognition in a plain backprop network. This doesn't do a huge number of iterations.

Definition at line 306 of file testTrainBasic.cpp.

Functions

Detailed Description

Function Documentation