Is anyone using scikit for NN?
Why/why not?
Why/why not?
[1] https://faiss.ai/ [2] https://github.com/facebookresearch/faiss
I mean, ignoring everything else, scikit has a much friendlier API.
- No saving checkpoints (can be crucial for large models who need alot of compute and time)
- No way to assign different activation functions to different layers
- No complex nodes like LSTM, GRU - No way to implement complex architectures like transformers, encoders etc
I also do not know if its even possible to use CUDA or any GPU with it.
[1] : https://scikit-learn.org/stable/modules/generated/sklearn.ne...
And AFAIK, there isn't GPU support, CPU performance is poor compared to GPU execution.
Skorch: https://github.com/skorch-dev/skorch
tf.keras.wrappers.scikit_learn: https://www.tensorflow.org/api_docs/python/tf/keras/wrappers...
AFAIU, there are not Yellowbrick visualizers for PyTorch or TensorFlow; though PyTorch abd TensorFlow work with TensorBoard for visualizing CFG execution.
> Many machine learning libraries implement the scikit-learn `estimator API` to easily integrate alternative optimization or decision methods into a data science workflow. Because of this, it seems like it should be simple to drop in a non-scikit-learn estimator into a Yellowbrick visualizer, and in principle, it is. However, the reality is a bit more complicated.
> Yellowbrick visualizers often utilize more than just the method interface of estimators (e.g. `fit()` and `predict()`), relying on the learned attributes (object properties with a single underscore suffix, e.g. `coef_`). The issue is that when a third-party estimator does not expose these attributes, truly gnarly exceptions and tracebacks occur. Yellowbrick is meant to aid machine learning diagnostics reasoning, therefore instead of just allowing drop-in functionality that may cause confusion, we’ve created a wrapper functionality that is a bit kinder with it’s messaging.
Looks like there are Yellowbrick wrappers for XGBoost, CatBoost, CuML, and Spark MLib; but not for NNs yet. https://www.scikit-yb.org/en/latest/api/contrib/wrapper.html...
From the RAPIDS.ai CuML team: https://docs.rapids.ai/api/cuml/stable/ :
> cuML is a suite of fast, GPU-accelerated machine learning algorithms designed for data science and analytical tasks. Our API mirrors Sklearn’s, and we provide practitioners with the easy fit-predict-transform paradigm without ever having to program on a GPU.
> As data gets larger, algorithms running on a CPU becomes slow and cumbersome. RAPIDS provides users a streamlined approach where data is intially loaded in the GPU, and compute tasks can be performed on it directly.
CuML is not an NN library; but there are likely performance optimizations from CuDF and CuML that would accelerate performance of NNs as well.
Dask ML works with models with sklearn interfaces, XGBoost, LightGBM, PyTorch, and TensorFlow: https://ml.dask.org/ :
> Scikit-Learn API
> In all cases Dask-ML endeavors to provide a single unified interface around the familiar NumPy, Pandas, and Scikit-Learn APIs. Users familiar with Scikit-Learn should feel at home with Dask-ML.
dask-labextension for JupyterLab helps to visualize Dask ML CFGs which call predictors and classifiers with sklearn interfaces: https://github.com/dask/dask-labextension
It is a pain to move in and out of scikit and write all these wrappers and converters. I would prefer to do more in one framework.
For example, doing hyperparameter optimization in pytorch using scikit can be a bit painful sometimes
> /? hierarchical automl "sklearn" site:github.com : https://www.google.com/search?q=hierarchical+automl+%22sklea...
https://westurner.github.io/hnlog/#comment-18798244
> Dask-ML works with {scikit-learn, xgboost, tensorflow, TPOT,}. ETL is your responsibility. Loading things into parquet format affords a lot of flexibility in terms of (non-SQL) datastores or just efficiently packed files on disk that need to be paged into/over in RAM. (Edit)
scale-scikit-learn https://examples.dask.org/machine-learning/scale-scikit-lear... -> dask.distributed parallel predication: https://examples.dask.org/machine-learning/parallel-predicti...
"Hyperparameter optimization with Dask" https://examples.dask.org/machine-learning/hyperparam-opt.ht...
> Sklearn.pipeline.Pipeline API: {fit(), transform(), predict(), score(),} https://scikit-learn.org/stable/modules/generated/sklearn.pi... : ```
decision_function(X) # Apply transforms, and decision_function of the final estimator
fit(X[, y]) # Fit the model
fit_predict(X[, y]) # Applies fit_predict of last step in pipeline after transforms.
fit_transform(X[, y]) # Fit the model and transform with the final estimator
get_params([deep]) # Get parameters for this estimator.
predict(X, *predict_params) # Apply transforms to the data, and predict with the final estimator
predict_log_proba(X) # Apply transforms, and predict_log_proba of the final estimator
predict_proba(X) # Apply transforms, and predict_proba of the final estimator
score(X[, y, sample_weight]) # Apply transforms, and score with the final estimator
score_samples(X) # Apply transforms, and score_samples of the final estimator.
set_params(**kwargs) # Set the parameters of this estimator
```
> https://docs.featuretools.com can also minimize ad-hoc boilerplate ETL / feature engineering :
>> Featuretools is a framework to perform automated feature engineering. It excels at transforming temporal and relational datasets into feature matrices for machine learning
From https://featuretools.alteryx.com/en/stable/guides/using_dask... :
> Creating a feature matrix from a very large dataset can be problematic if the underlying pandas dataframes that make up the entities cannot easily fit in memory. To help get around this issue, Featuretools supports creating Entity and EntitySet objects from Dask dataframes. A Dask EntitySet can then be passed to featuretools.dfs or featuretools.calculate_feature_matrix to create a feature matrix, which will be returned as a Dask dataframe. In addition to working on larger than memory datasets, this approach also allows users to take advantage of the parallel and distributed processing capabilities offered by Dask