GitHub - Karlxin/OK003: OpenKarlGeneralArtificialIntelligence

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 11 Commits
LBFGS2		LBFGS2
include		include
lib_win64		lib_win64
README.txt		README.txt
UFLDL_checkNNGradient.cpp		UFLDL_checkNNGradient.cpp
UFLDL_fmincg.cpp		UFLDL_fmincg.cpp
UFLDL_get_Cost_Grad.cpp		UFLDL_get_Cost_Grad.cpp
UFLDL_get_Cost_Grad_n.cpp		UFLDL_get_Cost_Grad_n.cpp
UFLDL_get_Cost_Grad_n_show.cpp		UFLDL_get_Cost_Grad_n_show.cpp
UFLDL_init_rand.cpp		UFLDL_init_rand.cpp
UFLDL_init_rand_n.cpp		UFLDL_init_rand_n.cpp
checkNNGradients.cpp		checkNNGradients.cpp
computeNumericalGradient.cpp		computeNumericalGradient.cpp
debugInitializeWeights.cpp		debugInitializeWeights.cpp
fmincg.cpp		fmincg.cpp
fmincg_3.cpp		fmincg_3.cpp
fmincg_n.cpp		fmincg_n.cpp
fminlbfgs.cpp		fminlbfgs.cpp
main.cpp		main.cpp
main.h		main.h
nnCostFunction.cpp		nnCostFunction.cpp
nnCostFunction_grad.cpp		nnCostFunction_grad.cpp
nnCostFunction_pair.cpp		nnCostFunction_pair.cpp
nnCostFunction_pair_3.cpp		nnCostFunction_pair_3.cpp
nnCostFunction_pair_n.cpp		nnCostFunction_pair_n.cpp
predict.cpp		predict.cpp
predict_3.cpp		predict_3.cpp
predict_n.cpp		predict_n.cpp
randInitializeWeights.cpp		randInitializeWeights.cpp
sigmoid.cpp		sigmoid.cpp
sigmoidGradient.cpp		sigmoidGradient.cpp

Repository files navigation

1.Target:
this project is about to create the general artificial intelligence

2.realizing:

3.problem: 1.the brain theory.we need more details about that.

4.reason:

5.todo: 1.create a reinforcement,supervised,unsupervised combination agent
2.reinforcement should have time and error punishment,adventure and correct reward,which is responsible for short term tactics.
this is somewhat the id.the subject.the humanity.the art.
3.supervised learning should take of longer term memory and a structure of value,which is responsible for engineering problem.
this is somewhat the ego.the project.the business.the technology.
4.unsupervised learning should take of the longest term,which is responsible for the general problem.
this is somewhat the superego.the object.the reason.the science.
6.phenomenon:

7.solutions:

8.exception:

postscript:
1.Hierarchical Temporal Memory

2.Sparse Distributed Representations

3.Spatial Pool
3.1.Delta States

4.Hebbian Reinforcement
4.1.Punishment
4.1.1.Time Punishment
4.1.2.Error Punishment
4.2.Reward
4.2.1.Adventure Reward
4.2.1.1.Learning Unknown
4.2.1.2.Exploring Unknown
4.2.2.Value Reward

5.Git
5.0.Avoiding Death
5.0.0.Old pattern may be important.We may just see the patterns instead of the reality.
5.1.Git Retrieve
5.1.0.Human synapses changing hard at ages,Nature give us death to avoid overfitting.
5.2.Git Merge
5.2.0.Merge let us combine two patterns group.