As years go by, Few Shot Learning (FSL) is becoming a hot topic not only in academic papers but also in production applications.

While a lot of researcher nowadays tend to publish their code on github, there is still no easy framework to get started with FSL. Especially when it comes to benchmarking existing models on personal datasets it is not always easy to find its path into each single repo. Not mentioning the Tensorflow/PyTorch issue.

This repo aims at filling this gap by providing a single entry-point for Few Shot Learning. It is deeply inspired by Keras because it shares the same philosophy:

It was developed with a focus on enabling fast experimentation. Being able to go from idea to result with the least possible delay is key to doing good research.

Thus this repo mainly relies on two of the main high-level python packages for data science: Keras and Pandas. While Pandas may not seem very useful for researchers working with static dataset, it becomes a strong backbone in production applications when you always need to tinker with your data.

Few-shot learning is a task consisting in classifying unseen samples into n classes (so called n way task) where each classes is only described with few (from 1 to 5 in usual benchmarks) examples.

Most of the state-of-the-art algorithms try to sort of learn a metric into a well suited (optimized) feature space. Thus deep networks usually first encode the base images into a feature space onto which a distance or similarity is learnt.

This similarity is meant to be used to later classify samples according to their relative distance, either in a pair-wise manner where the nearest support set samples is used to classify the query sample (Voronoi diagram) or in a more advanced classifier. Indeed, this philosophy is most commonly known as the kernel trick where the kernel is indeed the similarity learnt during training. Hence any kind of usual kernel based Machine Learning could potentially be plugged onto this learnt similarity.

There is no easy answer to the optimal choice of such a classifier in the feature space. This may depend on performance as well as one complexity and real application parameters. For instance if the support set is strongly imbalanced, you may not want to fit an advanced classifier onto it but rather use a raw nearest neighbor approach.

All these considerations lead to the need of a code architecture that will let you play with these parameters with your own data in order to take the best from them.

Amongst other, the Siamese Nets is usually known as the network from Koch et al. This algorithm learns a pair-wise similarity between images. More precisely it uses a densely connected layers on top of the difference between the two embeddings to predict 0 (different) or 1 (same).

In this repo we have called Siamese nets all the algorithms built within the same framework, i.e. choosing a backbone and a similarity to evaluate the embeddings. In this context the well known Protypical networks falls into the Siamese Nets frameworks and is available here as SiameseNets(head_model="ProtoNets").

This repos provides several tools for few-shot learning:

• Keras layers and models
• Keras sequences and Tensorflow datasets for training the models
• Notebooks with proven learning sequences

All these tools can be used all together or separately. One may want to stick with the keras model trained on regular numpy arrays, with or without callbacks. When designing more advanced keras.Sequence or tf.data.Dataset for training, it is advised (and some examples are provided) to use Pandas though it is not necessary at all.

We think that fast experimentation requires the good level of modularity. Modularity means that the flow of operations should be described as a sequence of operations with well defined interfaces. Furthermore you should be able to change or update any of these single operations without changing anything else. For these reasons we think that stacked Layers as well as stacked (chained) Pandas or tf.data operations are the right way of building ML pipelines. In this context we also rely on ImgAug for preprocessing the data.

Feel free to experiment and share your thought on this repo by contributing to it!

The notebooks section provides some examples. For instance, just run:

from keras_fsl.models import SiameseNets
from keras_fsl.datasets import omniglot
from keras_fsl.sequences.training.pairs import RandomPairsSequence

# %% Get data
train_set, test_set = omniglot.load_data()

# %% Update label columns to be able to mix alphabet during training
train_set = train_set.assign(label=lambda df: df.alphabet + '_' + df.label)
test_set = test_set.assign(label=lambda df: df.alphabet + '_' + df.label)

# %% Training
model = SiameseNets()
train_sequence = RandomPairsSequence(train_set, batch_size=16)
model.compile(optimizer='Adam', loss='binary_crossentropy')
model.fit_generator(train_sequence)