Skip to content

Latest commit

 

History

History
146 lines (84 loc) · 9.78 KB

CHANGELOG.md

File metadata and controls

146 lines (84 loc) · 9.78 KB

Changelog

All notable changes to this project will be documented in this file. See standard-version for commit guidelines.

1.5.2 (2024-06-19)

Bug Fixes

  • fixed a bug in the prediction function of MVE network. (fcada7d)

1.5.1 (2024-06-18)

Documentation

1.5.0 (2024-06-18)

Features

  • Made explicit mu and sigma layers for MVE network. (97165ab)

Documentation

1.4.0 (2024-06-17)

Features

  • Implemented Mean-variance network and cleaned up the dependencies. (448e865)

1.3.3 (2023-04-25)

Other

  • Added automatic formatting check for SciML compliance. (3fe21d8)

Documentation

  • Added a badge for the style compliance check. (8bdbee8)
  • Added a DOI Zenodo badge. (ea3f035)
  • small corrections (708f4b1)

1.3.2 (2022-07-26)

Documentation

  • Added a favicon as well. (b516ea5)
  • Made an EvidentialFlux logo based on the julia logo. :) (5a66ab6)

Other

  • Broke out the negative log likelihood calculation from the losses. (9e45c90)
  • Fixing styling inconsistencies with scimlstyle. (5259cd1)

1.3.0 (2022-07-19)

Features

  • Implemented the DER correction 🥶 (450a76f), closes #9
  • Updated regression example to produce something more useful and more inline with the paper. (8577d8e)

Documentation

  • Added classification image to readme. (9b3f64f)
  • Added documentation of the dirloss function. (e412f92)
  • Added more text regarding the classification entry. (afb03d1)
  • Added the documentation of the DIR layer. (736ea62)
  • Added the regression case to the README. (5fb3060)
  • Documented the Deep evidential classification. (f33f872)
  • Updated the example for the second regression case to use a power of 2 in the loss. (ec2026b)

1.2.0 (2022-07-12)

Features

  • Added the MultinomialDirichlet evidential distribution. (7436f16)

Documentation

  • Made some comments in the example. (e1f7d08)
  • Updated badges in README. (e2d9090)
  • Updated examples for classification. (daa32f4)

1.1.1 (2022-07-09)

Features

  • Added the evidence 🎓 function for the Deep Evidential Regression case. (5f92f07)

Documentation

  • Added documentation workflow to repo. (63051bd)
  • Added some information in the README again. (269fa89)
  • Added some information in the README again. (50bae43)
  • Fixed bug in badges. (3597864)
  • Rearranged order of the functions. (da95c62)
  • Section about the foundations of DER. (fee5f23)
  • Updated README with badge for documentation. (2ecfbd9)
  • Updated README with more badges. (e9a5614)

1.1.0 (2022-07-07)

Features

  • Added a full mini example evidential regression case. (9b3a649)
  • Added a predict function that dispatches on the last layer of a Flux Chain. (f81861a)
  • Added calculation function for uncertainty given a NIG network output. (829937c)
  • Added several tests for correctness. (b4f057d)
  • Added the NormalInverseGamma Dense layer and some tests. (c245be7)
  • Finished the evidential regression loss function. (3733528)
  • Speed up package by removing unneccesary dependencies. (d1e4631)
  • Updated the tests to include the full evidential regression loss. (76b9108)

Documentation

  • Added deployment of docs which is not working yet. (f3577d3)
  • Added documentation for the uncertainty functions (b5a2dc6)
  • Added Documenter infrastructure. (437ae08)
  • Added load path for local development. (2765605)
  • Documented the NIG loss functions as well as the aleatoric and epistemic uncertainty. (5f857e1)
  • Finished the documentation such that it can be generated. (6116f1b)
  • More documentation regardin the NIG distribution. (8282c91)