http://adelmanlab.org/easyNet/documentation/ 2026-04-07T17:24:54+00:00 http://adelmanlab.org/easyNet/documentation/ http://adelmanlab.org/easyNet/documentation/lib/tpl/dokuwiki/images/favicon.ico text/html 2016-03-25T14:25:24+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:bias_layer_impl&rev=1458915924&do=diff bias_layer_impl A layer_mem to be used with a vector_representation whose state and output is always scale. If a vector_representation layer is autoconnected to some keyed_representation layer, the resulting matrix_connection will have weights from the vector__ text/html 2016-03-25T13:12:42+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:clip_out&rev=1458911562&do=diff clip_out An output function that copies the state to the output, bounding the output unit-wise between clip_min and clip_max; i.e., if the state exceeds clip_max, the output is clip_max; if the state is less than clip_min, the output is clip_min, and otherwise the output is the same as the state. text/html 2016-03-25T13:12:42+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:fire_out&rev=1458911562&do=diff fire_out An output function that implements an exponential race model by treating the state levels (multiplied by rate) as the rate parameters of an exponential process for each unit. Thus the time between steps varies. When a step occurs, the output of the unit that has won this race is a copy of its state; all other outputs are zero. text/html 2016-03-25T13:12:42+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:iac_layer_extended_impl&rev=1458911562&do=diff iac_layer_extended_impl A transfer function that is used to mimic the synchronization bug in the SCM simulator for the decision layer. The “unshunted” channel is ignored and lateral inhibition is implemented internally --- with SCM's additional inhibition term multiplying by own state text/html 2016-03-25T13:12:42+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:iac_layer_impl&rev=1458911562&do=diff iac_layer_impl A transfer function that implements interactive activation. Parameters parameter default_value decay 0 rate 1 max_act 1 min_act 0 text/html 2016-03-25T13:12:42+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:integrate_transfer&rev=1458911562&do=diff integrate_transfer A transfer function that on each time step, increments the state by the product of the input and the duration of the time step. state(t+dt) = state(t) + input(t)*dt Recall that dt is not constant in all models; a layer with fire_out facet, for instance, has this effect on the model. text/html 2016-03-25T13:12:42+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:latch_transfer&rev=1458911562&do=diff latch_transfer Transfer function that implements a “latching” mechanism in one of two variants, depending on the value of the parameter keep_value. If keep_value is zero: All states start at zero. If the input to a unit is negative, set the state to text/html 2016-03-25T13:12:42+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:noise_layer_impl&rev=1458911562&do=diff noise_layer_impl An output function that calculates an output value from the state value by adding an independent standard normal N(0,1) variate. text/html 2016-03-25T14:26:36+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:scm_layer_impl&rev=1458915996&do=diff scm_layer_impl Implement SCM's word layer. Has three inputs: net_input, for inputs subject to normalization and “power law” transformation, “frequency_bias” for the biasing input of frequency, and “inhibitory” for inhibitory inputs not subject to normalization/_ text/html 2016-03-25T13:42:01+00:00 http://adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:layer:subtypes&rev=1458913321&do=diff Subtypes of layer Layer had four facet slots: layer_transfer associates inputs to state; layer_out associates state to output; layer_alloc allocates patterns for input/state/output; layer_mem manages memory, controlling, e.g., whether input, state and output differ, (relates to if layer___________________________________