v0.3.0

ExplainableAI v0.3.0

Diff since v0.2.0

Merged pull requests:

• Rename package to ExplainableAI (#44) (@adrhill)

v0.2.0

ExplainabilityMethods v0.2.0

Diff since v0.1.2

Closed issues:

• Faster LRP by dispatching on Dense and Conv layers (#22)
• Dispatch heatmap on method that was used to compute attribution (#23)
• Update docs with MNIST example and layer registration (#35)

Merged pull requests:

• Refactor heatmapping using symbols (#32) (@adrhill)
• Make LRP rules type inferable (#33) (@adrhill)
• Add Explanation type and dispatch heatmap on it. (#36) (@adrhill)
• Add inplace updating lrp! rule calls and reuse gradient code (#38) (@adrhill)
• Update docs with advanced LRP example (#40) (@adrhill)
• Update docs with LRP equations (#41) (@adrhill)
• Add outline to docs (#42) (@adrhill)
• Document LRP composites (#43) (@adrhill)

v0.1.2

ExplainabilityMethods v0.1.2

Diff since v0.1.1

Closed issues:

• Use DataAugmentation.jl for preprocessing in docs (#18)
• Add model checks for LRP (#25)

Merged pull requests:

• Use DataAugmentation.jl for preprocessing in docs (#20) (@adrhill)
• Restructure tests (#24) (@adrhill)
• Add LRP model checks and support unknown layers via AD fallback (#26) (@adrhill)
• Test and benchmark on VGG11 instead of VGG19 (#27) (@adrhill)
• Remove modify_layer (#29) (@adrhill)
• Fix tests and benchmarks on TestWrapper (#30) (@adrhill)
• Faster LRP rules on Dense layers (#31) (@adrhill)

v0.1.1

ExplainabilityMethods v0.1.1

Diff since v0.1.0

Closed issues:

• Add package benchmarking tools (#8)

Merged pull requests:

• Add package benchmarks (#15) (@adrhill)
• Fix safedivide (#16) (@adrhill)
• Fix RangeNormalizer (#17) (@adrhill)

v0.1.0

ExplainabilityMethods v0.1.0

Closed issues:

• Regression tests of all methods on VGG19 (#3)
• Regression tests of LRP rules on all common layers (#4)
• Add docs (#5)
• Fix randomness in gradients (#6)

Merged pull requests:

• Add LRP and gradient methods (#1) (@adrhill)
• Add reference tests on VGG-19 (#11) (@adrhill)
• Add layers and layer tests for LRP (#12) (@adrhill)
• Add documentation (#13) (@adrhill)