WojciechMigda / Tsetlini

Efficient parallelized implementation of Multilabel Classifier and Regressor Tsetlin Machines

https://wojciechmigda.github.io/Tsetlini/

WojciechMigda/Tsetlini Issues

Fix paths in `FindTsetlini.cmake`
Closed 9 months ago
Refresh `neither`
Closed 9 months ago
Refresh CI to address github deprecating ubuntu-18.04 image
Closed 9 months ago
Upgrade `actions/checkout`
Closed 2 years ago
Write regression algorithms unit tests
Closed 2 years ago
Split strong parameters into public and private subsets
Closed 2 years ago
Add extra optional parameters to `fit` and `partial_fit`
Updated 2 years ago1
Replace json factories invocation with variadic ones where possible
Closed 2 years ago
Rework `make_regressor_bitwise` into accepting variadic arguments instead of jsonized model parameters
Closed 2 years ago
Rework `make_regressor_classic` into accepting variadic arguments instead of jsonized model parameters
Closed 2 years ago
Rework `make_classifier_bitwise` into accepting variadic arguments instead of jsonized model parameters
Closed 2 years ago
[PILOT] Rework `make_classifier_classic` into accepting variadic arguments instead of jsonized model parameters
Closed 2 years ago
Implement `arg::maybe_extract`
Closed 2 years ago
Apply variadic arguments to public models' APIs
Closed 2 years ago
Introduce functions for extracting variadic arguments
Closed 2 years ago
Use `number_of_epochs_t` strong type at the public API level
Closed 2 years ago
Introduce `max_number_of_labels_t` strong type
Closed 2 years ago
Move `strong_type` library from `src` to `include`
Closed 2 years ago
Create Tsetlini's alias for `neither::Either`
Closed 2 years ago
Add extra check to bit vector feature data passed to bitwise algorithms
Closed 2 years ago
Use `ctest -j` to run tests in parallel in workflows
Closed 2 years ago
Add move assignment operators to Tsetlini estimators
Closed 2 years ago
Bytewise classifier training algorithm does not follow reference paper
Closed 2 years ago1
invalid
Closed 2 years ago
Fix integer rounding bias in exact `CoinTosser`
Closed 2 years ago
Remove unused code and data (regressor feedback)
Closed 2 years ago
Use `--fair-sched=yes` option with `valgrind`
Closed 2 years ago1
Replace references to 'signum' with 'polarity'
Closed 2 years ago
Use `with_pruning` instead of `for_predict` throughout the code
Closed 2 years ago
`sum_up_regressor_votes` with weighted clauses will integer overflow with sufficiently large weights
Closed 2 years ago
Weighted `sum_up_all_label_votes` has a type conversion bug
Closed 2 years ago
Use strong types
Closed 2 years ago1
Regressor can be created with an odd number of clauses
Closed 2 years ago
Move estimator state behind a pointer
Closed 2 years ago
Update `nlohmann/json` library
Closed 2 years ago
Extend existing valgrind and clang sanitizer checks
Closed 2 years ago1
Input response values validity is checked twice
Closed 2 years ago
Calling Regressor's `predict` API on a single sample throws exception
Closed 2 years ago
Regressor can be trained with input X matrix which has invalid number of features
Closed 2 years ago
Rename `predict_raw` as `decision_function`
Closed 2 years ago
Classifier and Regressor can be created with specificity parameter _s_ < 1.0
Closed 2 years ago
Fix number_of_regressor_clauses param validation
Closed 2 years ago
Input labels validity is checked twice
Closed 2 years ago1
Merge `fit_classifier_impl` and `fit_classifier_impl_T` into one function
Closed 2 years ago1
Classifier can be trained with input X matrix which has invalid number of features
Closed 2 years ago
Use fixed gtest version instead of latest master release
Closed 2 years ago
Create documentation
Updated 2 years ago
Add plantuml support in Sphinx docs
Closed 2 years ago1
Integrate Sphinx-based documentation
Closed 2 years ago1
Online training has a one-off error in input labels validation
Closed 2 years ago1