http://www.adelmanlab.org/easyNet/documentation/ 2026-04-24T00:02:06+00:00 http://www.adelmanlab.org/easyNet/documentation/ http://www.adelmanlab.org/easyNet/documentation/lib/tpl/dokuwiki/images/favicon.ico text/html 2016-03-25T15:03:13+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:arbitrary_conversion_connection_impl&rev=1458918193&do=diff arbitrary_conversion_connection_impl This connection uses a conversion between the source representation to the target representation to perform the connection. text/html 2016-03-25T15:03:27+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:best_above_threshold_connection_impl&rev=1458918207&do=diff best_above_threshold_connection A connection between two layers of the same representation using strength_patterns. The most active output dimensions of the source is identified. If this exceeds the threshold the output activity on this dimension only is copied to the target's input, with other dimensions of the target input zero. Otherwise, the target's input is all zero. text/html 2016-03-25T15:03:41+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:blocked_connection_impl&rev=1458918221&do=diff blocked_connection_impl Connects two layers with blocked_representations with matching numbers of blocks; corresponding blocks are attached with identical inner connections. Parameters Is a parameter_client that copies the names of the (first of the) inner connections, and delegates parameter setting to each of the inner connections (iteratively). text/html 2016-03-25T15:04:35+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:blocked_diag_connection_impl&rev=1458918275&do=diff blocked_diag_connection_impl Connects two layers with identical blocked_representations that use strength_patterns. Equivalent to a special matrix of matrices, in which the diagonal matrices have diagonal structure and off-diagonal matrices are zero. Also equivalent to a blocked__ text/html 2016-03-25T13:12:42+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:copy_connection_impl&rev=1458911562&do=diff copy_connection_impl.md Connection that copies the source layer output to the target layer input. text/html 2016-03-25T13:12:42+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:diag_connection_impl&rev=1458911562&do=diff diag_connection_impl Connection equivalent to a matrix whose diagonal elements are excitation and whose off-diagonal elements are inhibition. Parameters parameter default_value excitation 1 inhibition 0 text/html 2016-03-25T13:12:42+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:grep_connection_impl&rev=1458911562&do=diff grep_connection_impl Treats the source output pattern as a regex that is applied to the dimension names of the target input representation. Elements with matching dimension names are set to high_value and the rest to low_value. If the source output is empty, then the target input is set to zero. text/html 2016-03-25T15:07:44+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:lesioned_connection_impl&rev=1458918464&do=diff lesioned_connection_impl Does no operation on the target input. unlesion some_lconn unlesion Reinstate the stored connectionimpl of some_lconn and self-destruct the lesioned_connection_impl facet. text/html 2016-03-25T13:12:42+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:matrix_connection_impl&rev=1458911562&do=diff matrix_connection_impl Matrix multiplication is used to transform the source output to target input. details some_mconn details Print a human-readable list of matrix values. fill_from_dataframe some_mconn fill_from_dataframe some_df Use the dataframe object text/html 2016-03-25T13:12:42+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:prepost_diag_connection_impl&rev=1458911562&do=diff prepost_diag_connection_impl A connection for two layers with the same representation. First multiply the output of the source layer element-wise by a set of input multipliers; matrix multiply these in the same manner as diagconnectionimpl; and then multiply element-wise by a set of output multipliers and use the resultant values to set the input of the target layer. text/html 2016-03-25T13:12:42+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:selfweight_diag_connection_impl&rev=1458911562&do=diff selfweight_diag_connection_impl Connection for layers of the same representation. The input to the target layer is calculated as follows. For each element, sum an excitatory component and an inhibitory component. The excitatory component is excitation text/html 2016-03-25T15:09:15+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:sigmoid_diag_connection_impl&rev=1458918555&do=diff sigmoid_diag_connection_impl A version of diag_connection where inhibitory components are scaled with a sigmoid function. The excitation parameter is multiplies by the matches source value and added to inhibtion multiplied by the sum of a sigmoid function of each of the mismatching source values. The sigmoid function is text/html 2016-03-25T13:12:42+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:sparse_connection_impl&rev=1458911562&do=diff sparse_connection_impl Operates like a matrix (for multiplication) with a large number of inhibition values and a small number of excitation values. Notionally, for each dimension at the target layer, there is a small number of matching dimensions at the source layer, from which it receives excitation rather than inhibition. text/html 2016-03-25T15:11:06+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:spatial_hub_connection_impl&rev=1458918666&do=diff spatial_hub_connection_impl Connection to a keyed_representation target layer from a spatial_representation source layer. Sets the input of the target to the unnormalized (i.e., not divided by length) spatial coding match of the source layer output to the ideal spatial code given by converting the corresponding unit vector on the target layer to the spatial representation of the source. Some relevant parameters are parameters of the spatial representation.__ text/html 2016-03-25T15:12:31+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:staggered_blocked_connection_impl&rev=1458918751&do=diff staggered_blocked_connection_impl As blocked_connection, except each block is only connected if a delay (in the output time) has expired, which increases linearly with the block number. The minimum time at which the first block will be processed is given by text/html 2016-03-25T13:31:46+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:subtypes&rev=1458912706&do=diff Subtypes of connection best_above_threshold_connection = {best_above_threshold_connection_impl} copy_connection = {copy_connection_impl} grep_connection = {grep_connection_impl} prepostdiag_connection = {prepostdiag_connection_impl} selfweight_diag_connection = {selfweight_diag_connection_impl} sigmoid_diag_connection = {sigmoid_diag_connection_impl} spatial_hub_connection = {spatial_hub_connection_impl} spatial_tdf_connection = {spatial_tdf_connection_impl} staggered____________ text/html 2016-03-25T13:12:42+00:00 http://www.adelmanlab.org/easyNet/documentation/doku.php?id=lazynut_scripting_language:reference:connection:summate_connection_impl&rev=1458911562&do=diff summate_connection_impl Sum the activation on the source output, multiply by excitation and use this as a multiplier on the (localist) answer pattern to produce the target input. Parameters parameter default_value excitation 0 answer_pattern some_sumc answer_pattern apat