FEDOT Versions Save

Hi everyone!

We released a new major version of FEDOT - 0.5.0 It includes several major changes

The new version is available and can be imported via pip: pip install fedot==0.5.0

The most important changes:

• Preprocessing for tabular features and target variables improved dramatically.
• API is refactored and improved (presets, parameters, etc). The postfix "_tun" has been removed from the presets, so now you have to specify composer_params={'with_tuning': True} to set tuning. Important changes in preset names: light - best_quality, ultra_light - fast_train.
• Support for external optimisers is implemented.
• Zero-code console interface is implemented.
• Surrogate decision trees for pipeline interpretation are added.
• Custom model support is implemented.
• Better presets and models for time series forecasting (derivatives, polynomial models, cuts, etc)
• Better integration with FEDOT.Web
• Prototype for remote infrastructure support
• Evolutionary optimiser improved (stopping criterion, progress bar, better mutations, etc)
• Major and minor bugs are fixed.

We released a new major version of FEDOT - 0.4.1 It includes several large changes, features, and fixes.

The new version is available and can be imported via pip: pip install fedot==0.4.1

The most important changes:

• Major bugs fixed for evolutionary composing: we get rid of many annoying problems related to fitness evaluation and mutations
• Multi-variate time series forecasting improved
• Torch-based LSTM model added
• Encoding stage for categorial features implemented
• Docker containers updated, GPU example improved
• Export of pipelines improved
• Processing of hyperparameters improved
• API refactored

We released a new major version of FEDOT - 0.4.0 It includes several large changes, features, and fixes.

The new version is available and can be imported via pip: pip install fedot==0.4.0

The most important changes:

Infrastructure:

• Docker version added;
• GPU support added;
• Requirements become more flexible

Optimizer:

• Evolutionary optimizer generalized to allow the application to the custom non-ML tasks
• Mutation schemes improved for a more explainable evolution process;
• History saving extended

Time series:

• Cross-validation for the time series implemented;
• Sparse lagged transformation for time series implemented to improve performance;

Common:

• API updated and simplified;
• Processing of categorical features improved;
• Fixes and improvements for the hyperparameters tuning;
• The ‘chain’ term is replaced with ‘pipeline’ for better understandability.

Utilities:

• Sensitivity analysis improved

During the last month, we have merged several major features and fixed a bunch of bugs. Some of them are experimental and should be tested extensively in real-world cases. But we have tried our best and covered it with unit tests.

The new version (fedot == 0.3.1) is available and can be imported via pip.

The most important features:

• ML pipelines for multi-modal datasets
• Decompose operation in ML pipelines
• Cross-validation in Composer
• Add Memory and Time profilers
• Memory consumption improving

For details, see the post in repository: https://github.com/nccr-itmo/FEDOT/discussions/317

Hello everyone!

Our team finally has finished preparing a new major release of fedot == 0.3.0. Thanks to all dev team who was working on it! It is available and can be imported via pip: https://pypi.org/project/fedot/0.3.0/. The most important changes:

• Extended data operations and their automatic optimization

Previously, Fedot (Chain objects) allow one to automatically build ML pipelines including models, but data operations (like scaling or gap-filling) were embedded in the nodes and could be changed manually only. In the latest release we significantly refactored the core logic of the framework, thus data operations are fully supported as separate nodes. It can extend the overall search space of a suitable ML pipeline.

• New AutoML for time-series forecasting

Now Fedot supports not the only manual building of ML pipelines for time-series forecasting but also in an automated mode via Composer! Fedot allow one to build pipelines and forecast time-series for a given window size and forecasting length. Also, it is possible to use exogen variables for forecasting. To check all features, see examples in the repository.

Our early studies showed it is a promising approach that can improve AutoML field for time-series. We are actively working on the benchmarking of well-known SOTA frameworks for time-series forecasting and novel results will be published in a near future. Also, you can check our fresh preprint about gap filling in time-series using Fedot framework.

• Black-box optimization of ML pipeline hyperparameters

During the experiments, we found out that our previous version of tuning of hyperparameters seems to be ineffective (also it didn't work out for preprocessing nodes). Therefore, we significantly refactored the tuning module and it provides several schemas for black-box optimization of ML pipelines hyperparameters. For details, check tuning module sources and the examples.

• Multi-Objective AutoML for pipelines

Several months ago during the team discussion, we formulate a hypothesis: "Most of the AutoML frameworks are trying to maximize only one metric - prediction quality. But can we optimize several metrics (like pipeline complexity, for instance) simultaneously?" So we made research where evolutionary multi-objective optimization algorithms (like NSGA-II, SPEA-2) were adapted to the AutoML task. And it was concluded that it is a promising feature and we have integrated it into Fedot. The preprint is available, but also you can check the example how to use multi-objective optimization via Fedot API.

• New input data support for image classification

Later, we have announced that images will be supported in Fedot. And we made several changes in InputData and now pipelines for image classification can be built manually. We also added several CNN architectures and example of its usage. Composer should also work for image classification but we have not tested extensively this functionality yet.

Also, we have fixed a bunch of bugs and improved Fedot API.

Thanks to everyone who is following our progress! Any issues and user reports are welcomed. Cya!

Greetings to everyone who follows our team and FEDOT development progress!

Today, we released a new version of fedot == 0.2.1.

Here is the list of the main changes:

• Main API is updated. The basic 'how-to-use is available in the https://github.com/nccr-itmo/FEDOT/blob/master/notebooks/intro_to_automl.ipynb
• Support of the pandas dataframes is added
• Logging is improved
• The sensitivity analysis of the composite model (chain) structure is added. The description is available in https://fedot.readthedocs.io/en/latest/fedot/features/sensitivity_analysis.html

New version can be obtained using pip install fedot == 0.2.1

Оur team is very interested in any user feedback, the new issues are extremely welcomed! Thank you!

Greetings to everyone who follows our team and FEDOT development progress!

Last week, we released a new version of fedot == 0.2.0. A bunch of bugs in framework were fixed and merged to master (main) and release branches. Here is the list of the main changes:

• NLP tasks are now supported, a simple example of text classification were added (see here)
• The first version of fedot high-level API were implemented, see readme for the instructions
• Fixed several bugs with chain import/export
• Composer now should work correctly for time-series task
• Embedded visualization of composing and the resulted chains were improved, see the example here