Code

Dev/point process decoder design#53

Code

Closed

ncguilbeault wants to merge 131 commits intobonsai-rx:mainbonsai-rx/machinelearning:mainfrom ncguilbeault:dev/point-process-decoder-designncguilbeault/machinelearning:dev/point-process-decoder-designCopy head branch name to clipboard

Commits

Commits on Mar 12, 2025

• Added tensors library

  ncguilbeault

  committed
  bf99a53

  Copy full SHA for bf99a53

• Added arange function

  ncguilbeault

  committed
  19497a5

  Copy full SHA for 19497a5

• Added concat class

  ncguilbeault

  committed
  fd57f8c

  Copy full SHA for fd57f8c

• Added arange function

  ncguilbeault

  committed
  0d31457

  Copy full SHA for 0d31457

• Added linspace

  ncguilbeault

  committed
  0858c93

  Copy full SHA for 0858c93

• Added meshgrid

  ncguilbeault

  committed
  812be10

  Copy full SHA for 812be10

• Added ones

  ncguilbeault

  committed
  0d8f732

  Copy full SHA for 0d8f732

• Added zeros

  ncguilbeault

  committed
  c679dd4

  Copy full SHA for c679dd4

• Added device initialization

  ncguilbeault

  committed
  044f30e

  Copy full SHA for 044f30e

• Added ability to move tensor to device

  ncguilbeault

  committed
  9aeca78

  Copy full SHA for 9aeca78

• Added permute

  ncguilbeault

  committed
  daa2519

  Copy full SHA for daa2519

• Added reshape

  ncguilbeault

  committed
  05d88d5

  Copy full SHA for 05d88d5

• Added set

  ncguilbeault

  committed
  e07a7f7

  Copy full SHA for e07a7f7

• Updated csproj with opencv.net package

  ncguilbeault

  committed
  59ec8cd

  Copy full SHA for 59ec8cd

• Added concatenate class

  ncguilbeault

  committed
  b051999

  Copy full SHA for b051999

• Added convert data type

  ncguilbeault

  committed
  31b3981

  Copy full SHA for 31b3981

• Added create tensor method

  ncguilbeault

  committed
  78e7026

  Copy full SHA for 78e7026

• Added index method and updated set method

  ncguilbeault

  committed
  ac2f0cb

  Copy full SHA for ac2f0cb

• Defined tensor data types as subset of ScalarType

  ncguilbeault

  committed
  eb62f4b

  Copy full SHA for eb62f4b

• Added to array method

  ncguilbeault

  committed
  c36931c

  Copy full SHA for c36931c

• Added tensor data type helper

  ncguilbeault

  committed
  4f01250

  Copy full SHA for 4f01250

• Added methods to convert OpenCV types

  ncguilbeault

  committed
  f879960

  Copy full SHA for f879960

• Refactored to torch namespace instead of tensors

  ncguilbeault

  committed
  40615ab

  Copy full SHA for 40615ab

• Removed previous Bonsai.ML.Tensors directory and contents

  ncguilbeault

  committed
  1386ff3

  Copy full SHA for 1386ff3

• Updated solution to reflect change

  ncguilbeault

  committed
  0e820d7

  Copy full SHA for 0e820d7

• Moved Tensors namespace to main Torch namespace

  ncguilbeault

  committed
  f06a445

  Copy full SHA for f06a445

• Updated to reflect new namespace

  ncguilbeault

  committed
  980a03d

  Copy full SHA for 980a03d

• Removed unfinished classes

  ncguilbeault

  committed
  b3f3417

  Copy full SHA for b3f3417

• Updated to use common Bonsai.ML.Data project

  ncguilbeault

  committed
  df313d7

  Copy full SHA for df313d7

• Added additional overloads to process method

  ncguilbeault

  committed
  e75e2a9

  Copy full SHA for e75e2a9

• Updated to use common data tools

  ncguilbeault

  committed
  35f5308

  Copy full SHA for 35f5308

• Updated tensor data type and helper

  ncguilbeault

  committed
  c13c510

  Copy full SHA for c13c510

• Updated formatting

  ncguilbeault

  committed
  a7742ef

  Copy full SHA for a7742ef

• Removed bonsai core dependency in favor of bonsai.ml dependency

  ncguilbeault

  committed
  6bd2e66

  Copy full SHA for 6bd2e66

• Fixed bugs with create tensor method and updated to use string formatter

  ncguilbeault

  committed
  df6bd16

  Copy full SHA for df6bd16

• Added empty tensor creator

  ncguilbeault

  committed
  9b3c4f9

  Copy full SHA for 9b3c4f9

• Moved index helper to main library

  ncguilbeault

  committed
  2ced9dd

  Copy full SHA for 2ced9dd

• Moved opencv helper to main library

  ncguilbeault

  committed
  a2fa420

  Copy full SHA for a2fa420

• Removed opencv helper from helpers subfolder

  ncguilbeault

  committed
  05c12ce

  Copy full SHA for 05c12ce

• Updated with new index helper

  ncguilbeault

  committed
  57901ae

  Copy full SHA for 57901ae

• Updated with opencv helper

  ncguilbeault

  committed
  51ab15b

  Copy full SHA for 51ab15b

• Added process overload for generating on input

  ncguilbeault

  committed
  ac4b398

  Copy full SHA for ac4b398

• Added xml ignore tag on device

  ncguilbeault

  committed
  e69923a

  Copy full SHA for e69923a

• Updated to use shared module interface

  ncguilbeault

  committed
  9099cf9

  Copy full SHA for 9099cf9

• Modified to use Module interface

  ncguilbeault

  committed
  f76181e

  Copy full SHA for f76181e

• Removed unnecessary null string

  ncguilbeault

  committed
  befe81b

  Copy full SHA for befe81b

• Added a common interface

  ncguilbeault

  committed
  d4c4cc0

  Copy full SHA for d4c4cc0

• Added swap axes function

  ncguilbeault

  committed
  b9bacdd

  Copy full SHA for b9bacdd

• Added tile function

  ncguilbeault

  committed
  624c703

  Copy full SHA for 624c703

• Updated model loading and model forward procedure

  ncguilbeault

  committed
  2f799b7

  Copy full SHA for 2f799b7

• Added backward function for running online gradient descent with specified loss and optimization

  ncguilbeault

  committed
  637020e

  Copy full SHA for 637020e

• Added function to save model

  ncguilbeault

  committed
  65d8c70

  Copy full SHA for 65d8c70

• Changed name to indicate loading from an existing architecture

  ncguilbeault

  committed
  47dde6c

  Copy full SHA for 47dde6c

• Added descriptions and documentation

  ncguilbeault

  committed
  6d8442e

  Copy full SHA for 6d8442e

• Added some useful classes for linear algebra

  ncguilbeault

  committed
  f40aa4f

  Copy full SHA for f40aa4f

• Adding classes for creating tensor indexes

  ncguilbeault

  committed
  abb108b

  Copy full SHA for abb108b

• Updated MNIST model architecture with correct fully connected layer size

  ncguilbeault

  committed
  d8f19ff

  Copy full SHA for d8f19ff

• Added class to explicitly create a clone of a tensor

  ncguilbeault

  committed
  6331273

  Copy full SHA for 6331273

• Update to use correct width for an IplImage based on widthstep rather than width

  ncguilbeault

  committed
  4708507

  Copy full SHA for 4708507

• Explicitly use static torch.Tensor type for defining expressions

  ncguilbeault

  committed
  1926bcd

  Copy full SHA for 1926bcd

• Update set with process overloads to handle passing in tensor index

  ncguilbeault

  committed
  597265f

  Copy full SHA for 597265f

• Fixed incorrectly generating ones instead of zeros

  ncguilbeault

  committed
  5d2cb1e

  Copy full SHA for 5d2cb1e

• Updated to correctly parse colons in string

  ncguilbeault

  committed
  04a1f56

  Copy full SHA for 04a1f56

• Updated to use collection expressions

  ncguilbeault

  committed
  2dd1b91

  Copy full SHA for 2dd1b91

• Added documentation

  ncguilbeault

  committed
  f725c58

  Copy full SHA for f725c58

• Updated to use collection expressions

  ncguilbeault

  committed
  4f0776e

  Copy full SHA for 4f0776e

• Add process overload to initialize device on input

  ncguilbeault

  committed
  5b04e35

  Copy full SHA for 5b04e35

• Added documentation

  ncguilbeault

  committed
  2a6190d

  Copy full SHA for 2a6190d

• Added file name editor attribute to model path

  ncguilbeault

  committed
  9601460

  Copy full SHA for 9601460

• Updated to latest torchsharp version

  ncguilbeault

  committed
  506d8a1

  Copy full SHA for 506d8a1

• Update class name to reflect file name

  ncguilbeault

  committed
  ccfe400

  Copy full SHA for ccfe400

• Added documentation to package

  ncguilbeault

  committed
  ff8b463

  Copy full SHA for ff8b463

• Fixed issue with MNIST model not accepting num classes

  ncguilbeault

  committed
  2f4d6e3

  Copy full SHA for 2f4d6e3

• Made slight correction to GPU documentation

  ncguilbeault

  committed
  d17584f

  Copy full SHA for d17584f

• Modified torch module classes to be internel

  ncguilbeault

  committed
  49eef38

  Copy full SHA for 49eef38

• Reworked OpenCV <-> TorchSharp conversions to respect the garbage collector

  

  PathogenDavid

  authored and

  ncguilbeault

  committed
  1f063a9

  Copy full SHA for 1f063a9

• Removed unnecessary GCHandle usage (the lambda itself makes the anchor reachable instead), turned IplImage without backing data into an exception.

  Show description for 6cdd764
  

  PathogenDavid

  authored and

  ncguilbeault

  committed
  6cdd764

  Copy full SHA for 6cdd764

• Added point process model package

  ncguilbeault

  committed
  3a68d6b

  Copy full SHA for 3a68d6b

• Added get model function

  ncguilbeault

  committed
  fd4ea05

  Copy full SHA for fd4ea05

• Changed "IsDispose" to correct "IsDisposed" use of past tense

  ncguilbeault

  committed
  a911045

  Copy full SHA for a911045

• Added dispose method to model manager

  ncguilbeault

  committed
  4533762

  Copy full SHA for 4533762

• Updated comments/docs

  ncguilbeault

  committed
  2f2689a

  Copy full SHA for 2f2689a

• Added Bonsai.ML.Torch dependency

  ncguilbeault

  committed
  1f7fe53

  Copy full SHA for 1f7fe53

• Moved property group to top for consistency with other packages. Also updated the target frameworks to include net framework

  ncguilbeault

  committed
  3979a66

  Copy full SHA for 3979a66

• Updated dependency version

  ncguilbeault

  committed
  86a2174

  Copy full SHA for 86a2174

• Added documentation for point process decoder package

  ncguilbeault

  committed
  ff7140d

  Copy full SHA for ff7140d

• Updated point process model to include ignore no spikes parameter

  ncguilbeault

  committed
  63ce3be

  Copy full SHA for 63ce3be

• Updated point process decoder core package version

  ncguilbeault

  committed
  094964d

  Copy full SHA for 094964d

• Update decode method to allow modifying the ignore no spikes property in the clusterless likelihood calcultion

  ncguilbeault

  committed
  094173b

  Copy full SHA for 094173b

• Updated create model node to dispose model using the observable.finally method

  ncguilbeault

  committed
  b0fd4db

  Copy full SHA for b0fd4db

• Changed get model to source and add process function for no input

  ncguilbeault

  committed
  1254dfe

  Copy full SHA for 1254dfe

• Moved ignore no spikes property to likelihood parameters category

  ncguilbeault

  committed
  b1e0ca0

  Copy full SHA for b1e0ca0

• Added property descriptions

  ncguilbeault

  committed
  29058e2

  Copy full SHA for 29058e2

• Update package version

  ncguilbeault

  committed
  edeed96

  Copy full SHA for edeed96

• Update package dependency version

  ncguilbeault

  committed
  93c24ad

  Copy full SHA for 93c24ad

• Add kernel limit parameter to model

  ncguilbeault

  committed
  1b3d302

  Copy full SHA for 1b3d302

• Added load model class

  ncguilbeault

  committed
  2141c27

  Copy full SHA for 2141c27

• Added save model

  ncguilbeault

  committed
  8d680e6

  Copy full SHA for 8d680e6

• Removed try get call in create model

  ncguilbeault

  committed
  8307f73

  Copy full SHA for 8307f73

• Update suffix string format for datetime suffix type

  ncguilbeault

  committed
  6a64264

  Copy full SHA for 6a64264

• Added initial visualizers for conditional intensities and kernel estimates

  ncguilbeault

  committed
  0f5a1b4

  Copy full SHA for 0f5a1b4

• Make GetModel function public for visualizer

  ncguilbeault

  committed
  aa61162

  Copy full SHA for aa61162

• Removed kernel estimates visualizer

  ncguilbeault

  committed
  48e359f

  Copy full SHA for 48e359f

• Updated conditional intensities visualizer

  ncguilbeault

  committed
  454bd51

  Copy full SHA for 454bd51

• Adding all changes from altering dispose methods

  ncguilbeault

  committed
  65a0bb7

  Copy full SHA for 65a0bb7

• Moved toolstrip numeric up down class to seperate file

  ncguilbeault

  committed
  3d25d6b

  Copy full SHA for 3d25d6b

• Fixed issue with visualizers not displaying properly

  ncguilbeault

  committed
  ed29327

  Copy full SHA for ed29327

• Update heatmap series plot to make visualizer drop down public

  ncguilbeault

  committed
  5faf9a7

  Copy full SHA for 5faf9a7

• Update heatmap series to expose plot model and plot view publically.

  ncguilbeault

  committed
  e15854b

  Copy full SHA for e15854b

• Changed formatting. Used a lambda expression for on mouse view changed instead of explicit mouse event handler function.

  ncguilbeault

  committed
  1659700

  Copy full SHA for 1659700

• Make heatmap plot field public.

  ncguilbeault

  committed
  4327dbd

  Copy full SHA for 4327dbd

• Update decoder package to v0.3.0

  ncguilbeault

  committed
  21ae0ce

  Copy full SHA for 21ae0ce

• Update documentation for ignore no spikes property

  ncguilbeault

  committed
  bca52d8

  Copy full SHA for bca52d8

• Ensured ignore no spikes property is applied to both clusterless and poisson likelihood types

  ncguilbeault

  committed
  db968f3

  Copy full SHA for db968f3

• Added documentation to public get model function

  ncguilbeault

  committed
  a3661d2

  Copy full SHA for a3661d2

• Updated documentation for using Datetime suffix and used high resolution scheduler for extracting time information

  ncguilbeault

  committed
  b014433

  Copy full SHA for b014433

• Renamed ConditionalIntensitiesVisualizer to just IntensitiesVisualizer and made changes to display heatmaps correctly.

  ncguilbeault

  committed
  dbf2949

  Copy full SHA for dbf2949

• Added a density estimations visualizer

  ncguilbeault

  committed
  5af05b1

  Copy full SHA for 5af05b1

• Added a likelihood visualizer

  ncguilbeault

  committed
  dfd1c3e

  Copy full SHA for dfd1c3e

• Added a posterior visualizer

  ncguilbeault

  committed
  0514fcf

  Copy full SHA for 0514fcf

• Added a point2D overlay visualizer for posterior and likelihood mashup visualizers

  ncguilbeault

  committed
  e18af86

  Copy full SHA for e18af86

• Added Bonsai.Vision.Design package to project reference

  ncguilbeault

  committed
  4f8e962

  Copy full SHA for 4f8e962

• Updated heatmap series oxyplot base to enable setting x and y axis ranges from external class

  ncguilbeault

  committed
  aea32e2

  Copy full SHA for aea32e2

• Added a decoder visualizer interface

  ncguilbeault

  committed
  6825818

  Copy full SHA for 6825818

• Added a class which will automatically cycle through different oxyplot colors

  ncguilbeault

  committed
  cb7fc07

  Copy full SHA for cb7fc07

• Updated to use decoder visualizer interface and to use value min/max to set heatmap min/max from parent visualizer class

  ncguilbeault

  committed
  2e2f216

  Copy full SHA for 2e2f216

• Updated point overlay class to allow mashup visualizer class to handle setting the color

  ncguilbeault

  committed
  7d091d3

  Copy full SHA for 7d091d3

• Add SumAcrossBatch parameter to model constructor and decode function

  ncguilbeault

  committed
  d7b16c0

  Copy full SHA for d7b16c0

• Fixed issue with not being able to set the min/max heatmap values properly

  ncguilbeault

  committed
  0173558

  Copy full SHA for 0173558

• Updated PointProcessDecoder.Core package version

  ncguilbeault

  committed
  0ddb210

  Copy full SHA for 0ddb210

• Updated likelihood visualizer to place capacity property into visualizer properties dropdown menu

  ncguilbeault

  committed
  72f43aa

  Copy full SHA for 72f43aa