https://github.com/trungnt13/odin-ai

Orgainzed Digital Intelligent Network (O.D.I.N)
https://github.com/trungnt13/odin-ai

bayesian-methods deep-learning deep-neural-networks disentangled-representations disentanglement-learning factor-vae generative-model graph-algorithms image-processing machine-learning natural-language-processing probabilistic-graphical-models probabilistic-programming semi-supervised-learning speech-processing text-processing variational-autoencoder variational-autoencoders

Last synced: about 2 months ago
JSON representation

Orgainzed Digital Intelligent Network (O.D.I.N)

Host: GitHub
URL: https://github.com/trungnt13/odin-ai
Owner: trungnt13
License: mit
Created: 2016-06-25T22:25:31.000Z (almost 9 years ago)
Default Branch: master
Last Pushed: 2022-10-30T14:26:27.000Z (over 2 years ago)
Last Synced: 2025-03-15T20:33:42.813Z (about 2 months ago)
Topics: bayesian-methods, deep-learning, deep-neural-networks, disentangled-representations, disentanglement-learning, factor-vae, generative-model, graph-algorithms, image-processing, machine-learning, natural-language-processing, probabilistic-graphical-models, probabilistic-programming, semi-supervised-learning, speech-processing, text-processing, variational-autoencoder, variational-autoencoders
Language: Python
Homepage:
Size: 7.02 MB
Stars: 22
Watchers: 7
Forks: 11
Open Issues: 4
Metadata Files:
- Readme: README.rst
- License: LICENSE

Awesome Lists containing this project

README

        .. image:: https://readthedocs.org/projects/odin/badge/

    :target: http://odin0.readthedocs.org/en/latest/

O.D.I.N

=======

Organized Digital Intelligent Network (O.D.I.N)

O.D.I.N is a framework for building "Organized Digital Intelligent Networks".

End-to-end design, versatile, plug-n-play, minimized repetitive work

This repo contains the most comprehensive implementation of variational autoencoder and disentangled representation benchmark.

.. code-block:: python

  from odin.fuel import MNIST

  from odin.networks import get_networks

  from odin.bay.vi import VariationalAutoencoder

  ds = MNIST()

  train = ds.create_dataset(partition='train')

  # optimized architectures for MNIST

  networks = get_networks(ds, is_hierarchical=False, is_semi_supervised=False)

  # create the VAE

  vae = VariationalAutoencoder(**networks)

  vae.build(ds.full_shape)

  vae.fit(train, max_iter=10000)

TOC

---

1. `VAE`__

2. `Hierachical VAE`__

3. `Semi-supervised VAE`__

4. `Disentanglement Gym`__

5. `Faster Classical ML`__ (automatically select GPU implementation)

.. __: #variational-autoencoder-vae

.. __: #hierarchical-vae

.. __: #semi-supervised-vae

.. __: #disentanglement-gym

.. __: #fast-api-for-classical-ml

Variational Autoencoder (VAE)

-----------------------------

.. list-table::

   :widths: 30 80 25

   :header-rows: 1

   * - Model

     - Reference/Description

     - Implementation

   * - 1. Vanilla VAE

     - (Kingma et al. 2014). "Auto-Encoding Variational Bayes" [`Paper `_]

     - [`Code `_][`Example `_]

   * - 2. Beta-VAE

     - (Higgins et al. 2016). "beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework" [`Paper `_]

     - [`Code `_][`Example `_]

   * - 3. BetaGamma-VAE

     - Customized version of Beta-VAE, support re-weighing both reconstruction and regularization  ``\(\mathrm{ELBO}=\gamma \cdot E_q[log p(x|z)] - \beta \cdot KL(q(z|x)||p(z|x))\)``

     - [`Code `_][`Example `_]

   * - 4. Annealing VAE

     - (Sønderby et al. 2016) "Ladder variational autoencoder"

     - [`Code `_][`Example `_]

   * - 5. CyclicalAnnealing VAE

     - (Fu et al. 2019) "Cyclical Annealing Schedule: A Simple Approach to Mitigating KL Vanishing"

     - [`Code `_][`Example `_]

   * - 6. BetaTC-VAE

     - (Chen et al. 2019) "Isolating Sources of Disentanglement in Variational Autoencoders" (regularize the latents' Total Correlation)

     - [`Code `_][`Example `_]

   * - 7. Controlled Capacity Beta-VAE

     - (Burgess et al. 2018) "Understanding disentangling in beta-VAE"

     - [`Code `_][`Example `_]

   * - 8. FactorVAE

     - (Kim et al. 2018) "Disentangling by Factorising"

     - [`Code `_][`Example `_]

   * - 9. AuxiliaryVAE

     - (Maaløe et al. 2016) "Auxiliary Deep Generative Models"

     - [`Code `_][`Example `_]

   * - 10. HypersphericalVAE

     - (Davidson et al. 2018) "Hyperspherical Variational Auto-Encoders"

     - [`Code `_][`Example `_]

   * - 11. PowersphericalVAE

     - (De Cao et al. 2020) "The Power Spherical distribution"

     - [`Code `_][`Example `_]

   * - 12. DIPVAE

     - (Kumar et al. 2018) "Variational Inference of Disentangled Latent Concepts from Unlabeled Observations" (I - `only_mean=True`; II - `only_mean=False`)

     - [`Code `_][`Example `_]

   * - 13. InfoVAE

     - (Zhao et al. 2018) "infoVAE: Balancing Learning and Inference in Variational Autoencoders"

     - [`Code `_][`Example `_]

   * - 14. MIVAE

     - (Ducau et al. 2017) "Mutual Information in Variational Autoencoders" (max Mutual Information I(X;Z))

     - [`Code `_][`Example `_]

   * - 15. irmVAE

     - (Jing et al. 2020) "Implicit Rank-Minimizing Autoencoder" (Implicit Rank Minimizer)

     - [`Code `_][`Example `_]

   * - 16. ALDA

     - (Figurnov et al. 2018) "Implicit Reparameterization Gradients" (Amortized Latent Dirichlet Allocation - VAE with Dirichlet latents for topic modeling)

     - [`Code `_][`Example `_]

   * - 17. TwoStageVAE

     - (Dai et al. 2019) "Diagnosing and Enhancing VAE Models"

     - [`Code `_][`Example `_]

   * - 18. VampriorVAE

     - (Tomczak et al. 2018) "VAE with a VampPrior"

     - [`Code `_][`Example `_]

   * - 19. VQVAE

     - (Oord et al. 2017) "Neural Discrete Representation Learning"

     - [`Code `_][`Example `_]

Hierarchical VAE

----------------

.. list-table::

   :widths: 30 80 25

   :header-rows: 1

   * - Model

     - Reference/Description

     - Implementation

   * - 20. LadderVAE

     - (Sønderby et al. 2016) "Ladder variational autoencoder"

     - [`Code `_][`Example `_]

   * - 21. BidirectionalVAE

     - (Kingma et al. 2016) "Improved variational inference with inverse autoregressive flow" (Bidirectional inference hierarchical VAE)

     - [`Code `_][`Example `_]

   * - 22. ParallelVAE

     - (Zhao et al. 2017) "Learning Hierarchical Features from Generative Models" (Multiple latents connects encoder-decoder from bottom to top in parallel)

     - [`Code `_][`Example `_]

Semi-supervised VAE

-------------------

.. list-table::

   :widths: 30 80 25

   :header-rows: 1

   * - Model

     - Reference/Description

     - Implementation

   * - 23. Semi-supervised FactorVAE

     - Same as FactorVAE, but the discriminator also estimate the density of the labels and unlabeled data (like in semi-GAN)

     - [`Code `_][`Example `_]

   * - 24. MultiheadVAE

     - VAE has multiple decoders for different tasks

     - [`Code `_][`Example `_]

   * - 25. SkiptaskVAE

     - VAE has multiple tasks directly constrain the latents

     - [`Code `_][`Example `_]

   * - 26. ConditionalM2VAE

     - (Kingma et al. 2014) "Semi-supervised learning with deep generative models" [`Paper `_]

     - [`Code `_][`Example `_]

   * - 27. CCVAE (capture characteristic VAE)

     - (Joy et al. 2021) "Capturing label characteristics in VAEs" [`Paper `_]

     - [`Code `_][`Example `_]

   * - 28. SemafoVAE

     - (Trung et al. 2021) "The transitive information theory and its application to deep generative models" [`Paper `_]

     - [`Code `_][`Example `_]

Disentanglement Gym

-------------------

`DisentanglementGym `_: fast API for benchmarks on popular datasets and renowned disentanglement metrics.

Dataset support: `['shapes3d', 'dsprites', 'celeba', 'fashionmnist', 'mnist', 'cifar10', 'cifar100', 'svhn', 'cortex', 'pbmc', 'halfmoons']`

Metrics support:

* Correlation: 'spearman', 'pearson', 'lasso'

* BetaVAE score

* FactorVAE score

* Mutual Information Estimated

* MIG (Mutual Information Gap)

* SAP (Separated Attribute Prediction)

* RDS (relative disentanglement strength)

* DCI (Disentanglement, Completeness, Informativeness)

* FID (Frechet Inception Distance)

* Total Correlation

* Clustering scores: Adjusted Rand Index, Adjusted Mutual Info, Normalized Mutual Info, Silhouette score.

Fast API for classical ML

-------------------------

Automatically accelerated by RAPIDS.ai (i.e. automatically select GPU implementation if available)

Dimension Reduction

~~~~~~~~~~~~~~~~~~~

* t-SNE [`Code `_]

* UMAP [`Code `_]

* PCA, Probabilistic PCA, Supervised Probabilistic PCA, MiniBatch PCA, Randomize PCA [`Code `_]

* Probabilistic Linear Discriminant Analysis (PLDA) [`Code `_]

* iVector (GPU acclerated) [`Code `_]

GMM

~~~

* GMM classifier [`Code `_]

* Probabilistic embedding with GMM [`Code `_]

* Universal Background Model (GMM-Tmatrix) [`Code `_]

Clustering

~~~~~~~~~~

* KNN [`Code `_]

* KMeans [`Code `_]

* DBSCAN [`Code `_]

ecosyste.ms

Data

Tools

Indexes

Applications

Experiments

Awesome

https://github.com/trungnt13/odin-ai

Awesome Lists containing this project

README