aika-algorithm / aika Goto Github PK

AIKA is a new type of artificial neural network designed to more closely mimic the behavior of a biological brain and to bridge the gap to classical AI. A key design decision in the Aika network is to conceptually separate the activations from their neurons, meaning that there are two separate graphs. One graph consisting of neurons and synapses representing the knowledge the network has already acquired and another graph consisting of activations and links describing the information the network was able to infer about a concrete input data set. There is a one-to-many relation between the neurons and the activations. For example, there might be a neuron representing a word or a specific meaning of a word, but there might be several activations of this neuron, each representing an occurrence of this word within the input data set. A consequence of this decision is that we have to give up on the idea of a fixed layered topology for the network, since the sequence in which the activations are fired depends on the input data set. Within the activation network, each activation is grounded within the input data set, even if there are several activations in between. This means links between activations serve two purposes. On the one hand, they are used to sum up the synapse weights and, on the other hand they propagate the identity to higher level activations.

Home Page: https://aika.network

License: Apache License 2.0

Java 92.63% HTML 6.82% CSS 0.55%

nlp-machine-learning recurrent-neural-networks information-gain artificial-neural-network aika-network neural-symbolic-reasoning neural-symbolic neural-symbolic-paradigm entropy-coding binding-signal

aika's People

Contributors

Stargazers

Watchers

Forkers

kinow fvdaak benzei anujsrc sidra-z genjiluo stjordanis codeaudit ziponia vmpet3 guardian88 dougmota daoos chessengineus kishore7403 cyd3nt mayank-ramchandani wu0o0yu afi-22

aika's Issues

Python

Please implement the code also in the python language.

Optimizing Node.linkNeuronRelations

The Node.linkNeuronRelations method is called on every newly activated neuron to create the activation links to other activated neurons. This is done in both directions input and output. The problem here is that the number of incoming and outgoing synapses can be very large and the method will have to iterate over all of them to see if the linked neuron is activated. This works very well for a small number of synapses, but if the number is very large, it might be better to iterate over all activated neurons and see if there is a synapse between them. There is already a list of activated neurons in the Iteration object, but I don't think that it can be used here. It only contains activated neurons that are part of the final interpretation. This optimization, however, needs to iterate over all activated neurons. Therefore, another set of activated neurons is needed here.

Build tests failing with mvn clean test

Hi,

As commented on pull request #1, my environment seems to have problems running the tests. I am on branch master.

commit b7ab9fdbc6f1955fc9d6246cce49916bae51b527
Author: Lukas Molzberger <[email protected]>
Date:   Fri Jun 16 13:01:00 2017 +0200

    Organize imports

Running: mvn clean test with:

Apache Maven 3.3.9 (bb52d8502b132ec0a5a3f4c09453c07478323dc5; 2015-11-11T05:41:47+13:00)
Maven home: /opt/maven
Java version: 1.8.0_131, vendor: Oracle Corporation
Java home: /usr/lib/jvm/java-8-oracle/jre
Default locale: en_US, platform encoding: UTF-8
OS name: "linux", version: "4.4.0-79-generic", arch: "amd64", family: "unix"

Which is basically Ubuntu LTS with Oracle JDK 1.8. It fails with the following errors:

Tests run: 1, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.096 sec - in org.aika.network.SimplePatternMatchingTest

Results :

Failed tests: 
  PublicRangeTest.testPublicRange:101 null
  RecurrentNodeTest.testE:131 null
  RecurrentNodeTest.testFork:65 null
  RecurrentNodeTest.testOutputNode:200 null
  RecurrentNodeTest.testReverseDirection:166 null
Tests in error: 
  RecurrentNodeTest.testOutputNodeRD:272 NullPointer
  RecurrentNodeTest.testOverlappingClockAndTermSignals:341 NullPointer
  RecurrentNodeTest.testStartNode:396 NullPointer

Tests run: 61, Failures: 5, Errors: 3, Skipped: 0

[INFO] ------------------------------------------------------------------------
[INFO] BUILD FAILURE
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 9.672 s
[INFO] Finished at: 2017-06-16T23:38:45+12:00
[INFO] Final Memory: 19M/251M

There are a few places that inspecting the code in Eclipse, the IDE had a warning about possible NPE. It seems to happen indeed in some places in the log output.

testOverlappingClockAndTermSignals(org.aika.network.RecurrentNodeTest)  Time elapsed: 0 sec  <<< ERROR!
java.lang.NullPointerException: null
	at org.aika.network.RecurrentNodeTest.testOverlappingClockAndTermSignals(RecurrentNodeTest.java:341)

How to use

Quick question how do you use this? I cant seem to figure it out and theres no documentation coming with the code.

aika-algorithm / aika Goto Github PK

aika's People

Contributors

Stargazers

Watchers

Forkers

aika's Issues

Python

Optimizing Node.linkNeuronRelations

Build tests failing with mvn clean test

How to use

Recommend Projects

React

Vue.js

Typescript

TensorFlow

Django

Laravel

D3

Recommend Topics

javascript

web

server

Machine learning

Visualization

Game

Recommend Org

Facebook

Microsoft

Google

Alibaba

D3

Tencent