A flexible package for multimodal-deep-learning to combine tabular data with text and images using Wide and Deep models in Pytorch
Companion posts and tutorials: infinitoml
Experiments and comparison with
LightGBM: TabularDL vs LightGBM
The content of this document is organized as follows:
pytorch-widedeep is based on Google's Wide and Deep Algorithm,
adjusted for multi-modal datasets
In general terms,
pytorch-widedeep is a package to use deep learning with
tabular data. In particular, is intended to facilitate the combination of text
and images with corresponding tabular data using wide and deep models. With
that in mind there are a number of architectures that can be implemented with
just a few lines of code. The main components of those architectures are shown
in the Figure below:
The dashed boxes in the figure represent optional, overall components, and the
dashed lines/arrows indicate the corresponding connections, depending on
whether or not certain components are present. For example, the dashed,
blue-lines indicate that the
components are connected directly to the output neuron or neurons (depending
on whether we are performing a binary classification or regression, or a
multi-class classification) if the optional
deephead is not present.
Finally, the components within the faded-pink rectangle are concatenated.
Note that it is not possible to illustrate the number of possible
architectures and components available in
pytorch-widedeep in one Figure.
Therefore, for more details on possible architectures (and more) please, see
or the Examples folders and the notebooks there.
In math terms, and following the notation in the
paper, the expression for the architecture
deephead component can be formulated as:
Where 'W' are the weight matrices applied to the wide model and to the final activations of the deep models, 'a' are these final activations, and φ(x) are the cross product transformations of the original features 'x'. In case you are wondering what are "cross product transformations", here is a quote taken directly from the paper: "For binary features, a cross-product transformation (e.g., “AND(gender=female, language=en)”) is 1 if and only if the constituent features (“gender=female” and “language=en”) are all 1, and 0 otherwise".
While if there is a
deephead component, the previous expression turns
It is perfectly possible to use custom models (and not necessarily those in
the library) as long as the the custom models have an attribute called
output_dim with the size of the last layer of activations, so that
WideDeep can be constructed. Examples on how to use custom components can
be found in the Examples folder.
It is important to emphasize that each individual component,
deepimage, can be used independently and in
isolation. For example, one could use only
wide, which is in simply a
linear model. In fact, one of the most interesting functionalities
pytorch-widedeep would be the use of the
deeptabular component on
its own, i.e. what one might normally refer as Deep Learning for Tabular
pytorch-widedeep offers the following different models
for that component:
- Wide: a simple linear model where the nonlinearities are captured via cross-product transformations, as explained before.
- TabMlp: a simple MLP that receives embeddings representing the categorical features, concatenated with the continuous features, which can also be embedded.
- TabResnet: similar to the previous model but the embeddings are passed through a series of ResNet blocks built with dense layers.
- TabNet: details on TabNet can be found in TabNet: Attentive Interpretable Tabular Learning
Tabformer family, i.e. Transformers for Tabular data:
- TabTransformer: details on the TabTransformer can be found in TabTransformer: Tabular Data Modeling Using Contextual Embeddings.
- SAINT: Details on SAINT can be found in SAINT: Improved Neural Networks for Tabular Data via Row Attention and Contrastive Pre-Training.
- FT-Transformer: details on the FT-Transformer can be found in Revisiting Deep Learning Models for Tabular Data.
- TabFastFormer: adaptation of the FastFormer for tabular data. Details on the Fasformer can be found in FastFormers: Highly Efficient Transformer Models for Natural Language Understanding
- TabPerceiver: adaptation of the Perceiver for tabular data. Details on the Perceiver can be found in Perceiver: General Perception with Iterative Attention
And probabilistic DL models for tabular data based on Weight Uncertainty in Neural Networks:
- BayesianWide: Probabilistic adaptation of the
- BayesianTabMlp: Probabilistic adaptation of the
Note that while there are scientific publications for the TabTransformer, SAINT and FT-Transformer, the TabFasfFormer and TabPerceiver are our own adaptation of those algorithms for tabular data.
For details on these models (and all the other models in the library for the different data modes) and their corresponding options please see the examples in the Examples folder and the documentation.
Install using pip:
pip install pytorch-widedeep
Or install directly from github
pip install git+https://github.com/jrzaurin/pytorch-widedeep.git
# Clone the repository git clone https://github.com/jrzaurin/pytorch-widedeep cd pytorch-widedeep # Install in dev mode pip install -e .
Binary classification with the adult
DeepDense and defaults settings.
Building a wide (linear) and deep model with
import pandas as pd import numpy as np import torch from sklearn.model_selection import train_test_split from pytorch_widedeep import Trainer from pytorch_widedeep.preprocessing import WidePreprocessor, TabPreprocessor from pytorch_widedeep.models import Wide, TabMlp, WideDeep from pytorch_widedeep.metrics import Accuracy from pytorch_widedeep.datasets import load_adult df = load_adult(as_frame=True) df["income_label"] = (df["income"].apply(lambda x: ">50K" in x)).astype(int) df.drop("income", axis=1, inplace=True) df_train, df_test = train_test_split(df, test_size=0.2, stratify=df.income_label) # Define the 'column set up' wide_cols = [ "education", "relationship", "workclass", "occupation", "native-country", "gender", ] crossed_cols = [("education", "occupation"), ("native-country", "occupation")] cat_embed_cols = [ "workclass", "education", "marital-status", "occupation", "relationship", "race", "gender", "capital-gain", "capital-loss", "native-country", ] continuous_cols = ["age", "hours-per-week"] target = "income_label" target = df_train[target].values # prepare the data wide_preprocessor = WidePreprocessor(wide_cols=wide_cols, crossed_cols=crossed_cols) X_wide = wide_preprocessor.fit_transform(df_train) tab_preprocessor = TabPreprocessor( cat_embed_cols=cat_embed_cols, continuous_cols=continuous_cols # type: ignore[arg-type] ) X_tab = tab_preprocessor.fit_transform(df_train) # build the model wide = Wide(input_dim=np.unique(X_wide).shape, pred_dim=1) tab_mlp = TabMlp( column_idx=tab_preprocessor.column_idx, cat_embed_input=tab_preprocessor.cat_embed_input, continuous_cols=continuous_cols, ) model = WideDeep(wide=wide, deeptabular=tab_mlp) # train and validate trainer = Trainer(model, objective="binary", metrics=[Accuracy]) trainer.fit( X_wide=X_wide, X_tab=X_tab, target=target, n_epochs=5, batch_size=256, ) # predict on test X_wide_te = wide_preprocessor.transform(df_test) X_tab_te = tab_preprocessor.transform(df_test) preds = trainer.predict(X_wide=X_wide_te, X_tab=X_tab_te) # Save and load # Option 1: this will also save training history and lr history if the # LRHistory callback is used trainer.save(path="model_weights", save_state_dict=True) # Option 2: save as any other torch model torch.save(model.state_dict(), "model_weights/wd_model.pt") # From here in advance, Option 1 or 2 are the same. I assume the user has # prepared the data and defined the new model components: # 1. Build the model model_new = WideDeep(wide=wide, deeptabular=tab_mlp) model_new.load_state_dict(torch.load("model_weights/wd_model.pt")) # 2. Instantiate the trainer trainer_new = Trainer(model_new, objective="binary") # 3. Either start the fit or directly predict preds = trainer_new.predict(X_wide=X_wide, X_tab=X_tab)
Of course, one can do much more. See the Examples folder, the documentation or the companion posts for a better understanding of the content of the package and its functionalities.
How to Contribute
Check CONTRIBUTING page.
This library takes from a series of other libraries, so I think it is just fair to mention them here in the README (specific mentions are also included in the code).
TextProcessor class in this library uses the
Vocab. The code at
utils.fastai_transforms is a minor
adaptation of their code so it functions within this library. To my experience
Tokenizer is the best in class.
ImageProcessor class in this library uses code from the fantastic Deep
Learning for Computer
(DL4CV) book by Adrian Rosebrock.