marcospeng/carefree-learn

A minimal Automatic Machine Learning (AutoML) solution for tabular datasets based on PyTorch

carefree-learn

carefree-learn is a minimal Automatic Machine Learning (AutoML) solution for tabular datasets based on PyTorch.

Why carefree-learn?

carefree-learn

• Provides a scikit-learn-like interface with much more 'carefree' usages, including:
  • Automatically deals with data pre-processing.
  • Automatically handles datasets saved in files (.txt, .csv).
  • Supports Distributed Training, which means hyper-parameter tuning can be very efficient in carefree-learn.
• Includes some brand new techniques which may boost vanilla Neural Network (NN) performances on tabular datasets, including:
  • TreeDNN with Dynamic Soft Pruning, which makes NN less sensitive to hyper-parameters.
  • Deep Distribution Regression (DDR), which is capable of modeling the entire conditional distribution with one single NN model.
• Supports many convenient functionality in deep learning, including:
  • Early stopping.
  • Model persistence.
  • Learning rate schedulers.
  • And more...
• Full utilization of the WIP ecosystem cf*, such as:
  • carefree-toolkit: provides a lot of utility classes & functions which are 'stand alone' and can be leveraged in your own projects.
  • carefree-data: a lightweight tool to read -> convert -> process ANY tabular datasets. It also utilizes cython to accelerate critical procedures.

From the above, it comes out that carefree-learn could be treated as a minimal Automatic Machine Learning (AutoML) solution for tabular datasets when it is fully utilized. However, this is not built on the sacrifice of flexibility. In fact, the functionality we've mentioned are all wrapped into individual modules in carefree-learn and allow users to customize them easily.

Installation

carefree-learn requires Python 3.6 or higher.

Pre-Installing PyTorch

Please refer to PyTorch, and it is highly recommended to pre-install PyTorch with conda.

pip installation

After installing PyTorch, installation of carefree-learn would be rather easy:

• Tips: if you pre-installed PyTorch with conda, remember to activate the corresponding environment!

git clone https://github.com/carefree0910/carefree-learn.git
cd carefree-learn
pip install -e .

AutoML

carefree-learn provides cflearn.Auto API for out-of-the-box usages.

TL; DR

import cflearn

from cfdata.tabular import *

# prepare iris dataset
iris = TabularDataset.iris()
iris = TabularData.from_dataset(iris)
# split 10% of the data as validation data
split = iris.split(0.1)
train, valid = split.remained, split.split
x_tr, y_tr = train.processed.xy
x_cv, y_cv = valid.processed.xy
data = x_tr, y_tr, x_cv, y_cv


if __name__ == '__main__':
    # standard usage
    fcnn = cflearn.make().fit(*data)

    # 'overfit' validation set
    # * `clf` indicates this is a classification task
    # * for regression tasks, use `reg` instead
    auto = cflearn.Auto("clf").fit(*data, num_jobs=2)

    # estimate manually
    predictions = auto.predict(x_cv)
    print("accuracy:", (y_cv == predictions).mean())

    # estimate with `cflearn`
    cflearn.estimate(
        x_cv,
        y_cv,
        pipelines=fcnn,
        other_patterns={"auto": auto.pattern},
    )

Then you will see something like this:

================================================================================================================================
|        metrics         |                       acc                        |                       auc                        |
--------------------------------------------------------------------------------------------------------------------------------
|                        |      mean      |      std       |     score      |      mean      |      std       |     score      |
--------------------------------------------------------------------------------------------------------------------------------
|          auto          | -- 1.000000 -- | -- 0.000000 -- | -- 1.000000 -- | -- 1.000000 -- | -- 0.000000 -- | -- 1.000000 -- |
--------------------------------------------------------------------------------------------------------------------------------
|          fcnn          |    0.933333    | -- 0.000000 -- |    0.933333    |    0.993333    | -- 0.000000 -- |    0.993333    |
================================================================================================================================

Explained

cflearn.Auto.fit will run through the following steps:

1. define the model space automatically (or manually)
2. fetch pre-defined hyper-parameters search space of each model from OptunaPresetParams.
3. leverage optuna with cflearn.optuna_tune to perform hyper-parameters optimization.
4. use searched hyper-parameters to train each model multiple times (separately).
5. ensemble all trained models (with cflearn.Ensemble.stacking).
6. record all these results to corresponding attributes.

So after cflearn.Auto.fit, we can perform visualizations provided by optuna easily:

export_folder = "iris_vis"
auto.plot_param_importances("fcnn", export_folder=export_folder)
auto.plot_intermediate_values("fcnn", export_folder=export_folder)

Examples

• Here are some real life examples.
• For detailed information, please visit the documentation.

Quick Start

import cflearn
from cfdata.tabular import TabularDataset

x, y = TabularDataset.iris().xy
m = cflearn.make().fit(x, y)
# Make label predictions
m.predict(x)
# Make probability predictions
m.predict_prob(x)
# Estimate performance
cflearn.estimate(x, y, pipelines=m)

Then you will see something like this:

================================================================================================================================
|        metrics         |                       acc                        |                       auc                        |
--------------------------------------------------------------------------------------------------------------------------------
|                        |      mean      |      std       |     score      |      mean      |      std       |     score      |
--------------------------------------------------------------------------------------------------------------------------------
|          fcnn          |    0.946667    |    0.000000    |    0.946667    |    0.993200    |    0.000000    |    0.993200    |
================================================================================================================================

It is also worth mentioning that carefree-learn models can be saved easily, into a zip file!

For example, a cflearn^_^fcnn.zip file will be created with one line of code:

cflearn.save(m)

Of course, loading carefree-learn models are easy too!

m = cflearn.load()
# You will see exactly the same result as above!
cflearn.estimate(x, y, pipelines=m)

carefree-learn can also easily fit / predict / estimate directly on files (file-in, file-out). Suppose we have an 'xor.txt' file with following contents:

0,0,0
0,1,1
1,0,1
1,1,0

Then carefree-learn can be utilized with only one line of code:

delim refers to 'delimiter', and skip_first refers to whether skipping the first line or not.

Please refer to carefree-data if you're interested in more details.

m = cflearn.make(delim=",", skip_first=False).fit("xor.txt", x_cv="xor.txt")
cflearn.estimate("xor.txt", pipelines=m)

After which you will see something like this:

================================================================================================================================
|        metrics         |                       acc                        |                       auc                        |
--------------------------------------------------------------------------------------------------------------------------------
|                        |      mean      |      std       |     score      |      mean      |      std       |     score      |
--------------------------------------------------------------------------------------------------------------------------------
|          fcnn          |    1.000000    |    0.000000    |    1.000000    |    1.000000    |    0.000000    |    1.000000    |
================================================================================================================================

When we fit from files, we can predict on either files or lists:

print(m.predict([[0, 0]]))   # [[0]]
print(m.predict([[0, 1]]))   # [[1]]
print(m.predict("xor.txt"))  # [ [0] [1] [1] [0] ]

Distributed

In carefree-learn, Distributed Training doesn't mean training your model on multiple GPUs or multiple machines, because carefree-learn focuses on tabular datasets (or, structured datasets) which are often not as large as unstructured datasets. Instead, Distributed Training in carefree-learn means training multiple models at the same time. This is important because:

• Deep Learning models suffer from randomness, so we need to train multiple models with the same algorithm and calculate the mean / std of the performances to estimate the algorithm's capacity and stability.
• Ensemble these models (which are trained with the same algorithm) can boost the algorithm's performance without making any changes to the algorithm itself.
• Parameter searching will be easier & faster.

import cflearn
from cfdata.tabular import TabularDataset

# It is necessary to wrap codes under '__main__' on WINDOWS platform when running distributed codes
if __name__ == '__main__':
    x, y = TabularDataset.iris().xy
    # Notice that 3 fcnn were trained simultaneously with this line of code
    results = cflearn.repeat_with(x, y, num_repeat=3, num_jobs=0)
    patterns = results.patterns["fcnn"]
    # And it is fairly straight forward to apply stacking ensemble
    ensemble = cflearn.Ensemble.stacking(patterns)
    patterns_dict = {"fcnn_3": patterns, "fcnn_3_ensemble": ensemble}
    cflearn.estimate(x, y, metrics=["acc", "auc"], other_patterns=patterns_dict)

Then you will see something like this:

================================================================================================================================
|        metrics         |                       acc                        |                       auc                        |
--------------------------------------------------------------------------------------------------------------------------------
|                        |      mean      |      std       |     score      |      mean      |      std       |     score      |
--------------------------------------------------------------------------------------------------------------------------------
|         fcnn_3         |    0.937778    |    0.017498    |    0.920280    | -- 0.993911 -- |    0.000274    |    0.993637    |
--------------------------------------------------------------------------------------------------------------------------------
|    fcnn_3_ensemble     | -- 0.953333 -- | -- 0.000000 -- | -- 0.953333 -- |    0.993867    | -- 0.000000 -- | -- 0.993867 -- |
================================================================================================================================

You might notice that the best results of each column is 'highlighted' with a pair of '--'.

Hyper Parameter Optimization (HPO)

import cflearn
from cfdata.tabular import TabularDataset
 
if __name__ == '__main__':
    x, y = TabularDataset.iris().xy
    # Bayesian Optimization (BO) will be used as default
    hpo = cflearn.tune_with(
        x, y,
        task_type="clf",
        num_repeat=2, num_parallel=0, num_search=10
    )
    # We can further train our model with the best hyper-parameters we've obtained:
    m = cflearn.make(**hpo.best_param).fit(x, y)
    cflearn.estimate(x, y, pipelines=m)

Then you will see something like this:

~~~  [ info ] Results
================================================================================================================================
|        metrics         |                       acc                        |                       auc                        |
--------------------------------------------------------------------------------------------------------------------------------
|                        |      mean      |      std       |     score      |      mean      |      std       |     score      |
--------------------------------------------------------------------------------------------------------------------------------
|        0659e09f        |    0.943333    |    0.016667    |    0.926667    |    0.995500    |    0.001967    |    0.993533    |
--------------------------------------------------------------------------------------------------------------------------------
|        08a0a030        |    0.796667    |    0.130000    |    0.666667    |    0.969333    |    0.012000    |    0.957333    |
--------------------------------------------------------------------------------------------------------------------------------
|        1962285c        |    0.950000    |    0.003333    |    0.946667    |    0.997467    |    0.000533    |    0.996933    |
--------------------------------------------------------------------------------------------------------------------------------
|        1eb7f2a0        |    0.933333    |    0.020000    |    0.913333    |    0.994833    |    0.003033    |    0.991800    |
--------------------------------------------------------------------------------------------------------------------------------
|        4ed5bb3b        |    0.973333    |    0.013333    |    0.960000    |    0.998733    |    0.000467    |    0.998267    |
--------------------------------------------------------------------------------------------------------------------------------
|        5a652f3c        |    0.953333    | -- 0.000000 -- |    0.953333    |    0.997400    |    0.000133    |    0.997267    |
--------------------------------------------------------------------------------------------------------------------------------
|        82c35e77        |    0.940000    |    0.020000    |    0.920000    |    0.995467    |    0.002133    |    0.993333    |
--------------------------------------------------------------------------------------------------------------------------------
|        a9ef52d0        | -- 0.986667 -- |    0.006667    | -- 0.980000 -- | -- 0.999200 -- | -- 0.000000 -- | -- 0.999200 -- |
--------------------------------------------------------------------------------------------------------------------------------
|        ba2e179a        |    0.946667    |    0.026667    |    0.920000    |    0.995633    |    0.001900    |    0.993733    |
--------------------------------------------------------------------------------------------------------------------------------
|        ec8c0837        |    0.973333    | -- 0.000000 -- |    0.973333    |    0.998867    |    0.000067    |    0.998800    |
================================================================================================================================

~~~  [ info ] Best Parameters
----------------------------------------------------------------------------------------------------
acc  (a9ef52d0) (0.986667 ± 0.006667)
----------------------------------------------------------------------------------------------------
{'optimizer': 'rmsprop', 'optimizer_config': {'lr': 0.005810863965757382}}
----------------------------------------------------------------------------------------------------
auc  (a9ef52d0) (0.999200 ± 0.000000)
----------------------------------------------------------------------------------------------------
{'optimizer': 'rmsprop', 'optimizer_config': {'lr': 0.005810863965757382}}
----------------------------------------------------------------------------------------------------
best (a9ef52d0)
----------------------------------------------------------------------------------------------------
{'optimizer': 'rmsprop', 'optimizer_config': {'lr': 0.005810863965757382}}
----------------------------------------------------------------------------------------------------

~~  [ info ] Results
================================================================================================================================
|        metrics         |                       acc                        |                       auc                        |
--------------------------------------------------------------------------------------------------------------------------------
|                        |      mean      |      std       |     score      |      mean      |      std       |     score      |
--------------------------------------------------------------------------------------------------------------------------------
|          fcnn          |    0.980000    |    0.000000    |    0.980000    |    0.998867    |    0.000000    |    0.998867    |
================================================================================================================================

You might notice that:

• The final results obtained by HPO is even better than the stacking ensemble results mentioned above.
• We search for optimizer and lr as default. In fact, we can manually passed params into cflearn.tune_with. If not, then carefree-learn will execute following codes:

from cftool.ml.param_utils import *

params = {
    "optimizer": String(Choice(values=["sgd", "rmsprop", "adam"])),
    "optimizer_config": {
        "lr": Float(Exponential(1e-5, 0.1))
    }
}

It is also worth mention that we can pass file datasets into cflearn.tune_with as well. See tests/usages/test_basic.py for more details.

Citation

If you use carefree-learn in your research, we would greatly appreciate if you cite this library using this Bibtex:

@misc{carefree-learn,
  year={2020},
  author={Yujian He},
  title={carefree-learn, a minimal Automatic Machine Learning (AutoML) solution for tabular datasets based on PyTorch},
  howpublished={\url{https://https://github.com/carefree0910/carefree-learn/}},
}

License

carefree-learn is MIT licensed, as found in the LICENSE file.