Optimization of the interpretation search using an upper bound on the interpretation weights. Rewrite of the conversion of synapse weights to logic nodes. The APIs allow to implement heuristics when deciding which synapses should be created or whichĮxperimental support for text generation.
#AIKA ONLINE TCOIN GENERATOR CODE#
Removed some old experimental training code and provided two APIs for training and pattern discovery. Suppressed if they are conflicting with other activations. Simplification: Activations are now only added during processing never removed. Optimization of the checkSelfReferencing function.įixes for the training and pattern discovery functions. Optimization for the search for the best interpretation. The bias delta value in a neuron input is now an absolute value.īug fixes, code cleanups, code readability improvements, lambda expression usage, convenience functions. Memory optimization: Disjunctive synapses are now stored on the input neuron side. Refactoring of the range matching within synapses. Optimization and simplification of the interpretation search. The debugging output is much more detailed now.ĪPI cleanups: Input -> Synapse.Builder, Activation.Builder The search is now iterative to prevent stack overflows. Refactoring of the interpretation search. Moved the activation linking and activation selection code to separate classes.Ĭaching of partially computed states in the neural network during the interpretation search. Simplified interpretation handling by removing the InterpretationNode class and moving the remaining logic to the Activation class. Work on an syllable identification experiment based on the meta network implementation.
#AIKA ONLINE TCOIN GENERATOR FREE#
On the distance of the activation ranges.Įxample implementation of a context free grammar.Įxample implementation for co-reference resolution. It allows to model a weakening signal depending Introduced an optional distance function to synapses. Removed the norm term from the interpretation objective function. The input activations of two synapses, but instead of the ranges the dependency relations of these Range relations compare the inputĪctivation range of a given synapse with that of the linked synapse. Two types of relations areĬurrently supported, range relations and instance relations. Input activation 1 equals to the begin position of input activation 2. Now it is possible to explicitly model relations like: The end position of Previously the relation between synapses was implicitly modeled Optimization of the interpretation search. Passive neurons act basically like callback functions. Passive neurons are only evaluated if the connected output neuron requires it. Use case the positions are not known in advance and need to be computed during processing. This feature is required for text generation. The range positions are now optionally variable. The synapse bias is now used to decide whether a synapse is conjunctive or disjunctive.ĪPI Refactoring: Synapse relations are now established through a separate builder class. Now activations can possess an arbitrary number of positions. Previously, an activation had a range with a begin and an end position. The Range class has now been replaced with by a slots concept. Refactoring of the relation architecture. Major simplification of the interpretation search. Implementation to compute the average activation of an activation object over all interpretation options. Replaced the search for the optimal interpretations with branching of activations. Replaced the relations between synapses with the concept of pattern part neurons. Removed the node lattice from the algorithm The implementation of the math is now declarative instead
Replaced the hard coded math with a dynamic field concept that allows to arbitrarily propagateĬhanges through pipeline like connections. Training gradient computed from an objective function based on the entropy of the neuron and the
Rewrite of the algorithm to use binding-signals for the linking process. JavaDoc v2.0.2-alpha Release Notes v2.0.2-alpha () If you would like to get involved, please contact me. Every contribution is welcome and needed to make AIKA better.
0 kommentar(er)
