New metric proposed to accurately measure disentanglement.
problem Lack of theoretical guarantees and consistent correlation for existing disentanglement metrics.
method Analysis and creation of a new metric.
result Proposed metric satisfies two desirable properties for disentanglement.
We analyze disentangled representations under a causal generative process, proposing new metrics and datasets.
problem Addressing fairness and interpretability through disentangled representations with a causal perspective.
method Work under a causal generative process, proposing new metrics and datasets to study disentanglement.
result Proposed metrics capture the desiderata of disentangled causal process.
New metric for disentangling multivariate representations, accounting for more complex entanglements.
problem Current disentanglement metrics fail to detect entanglements involving more than two variables.
method Partial Information Decomposition framework to analyze information sharing and propose a new disentanglement metric.
result The proposed metric correctly identifies entanglements in high-dimensional spaces.
We make two theoretical contributions to disentanglement learning by (a) defining precise semantics of disentangled representations, and (b) establishing robust metrics for evaluation. First, we characterize the concept "disentangled representations" used in supervised and unsupervised methods along three dimensions-in…
Study evaluates unsupervised disentanglement methods on a toy dataset.
problem Lack of clear disentanglement metrics capturing independent features.
method Empirical evaluation of six unsupervised disentanglement methods on MPI3D dataset.
result Beta-TCVAE outperforms other methods in metrics, but not in disentanglement quality.
We define and address the problem of unsupervised learning of disentangled representations on data generated from independent factors of variation. We propose FactorVAE, a method that disentangles by encouraging the distribution of representations to be factorial and hence independent across the dimensions. We show tha…
Paper introduces Wasserstein total correlation for disentangled representation learning.
problem Learning disentangled representations from data.
method Adversarial training of a critic to estimate Wasserstein total correlation in variational and Wasserstein autoencoders.
result Proposed method achieves comparable disentanglement performance with less reconstruction loss.
This paper critically examines unsupervised disentangled representation learning, revealing challenges and limitations.
problem The difficulty of unsupervised learning of disentangled representations and the challenges in evaluation metrics.
method Theoretical analysis and a large-scale experimental study covering 8 datasets and 14000 models.
result Well-disentangled models cannot be identified without supervision, and different evaluation metrics disagree on what constitutes disentanglement.
Disentangled representations, where the higher level data generative factors are reflected in disjoint latent dimensions, offer several benefits such as ease of deriving invariant representations, transferability to other tasks, interpretability, etc. We consider the problem of unsupervised learning of disentangled rep…
This paper improves disentanglement in VAEs by progressively learning hierarchical representations.
problem Compromised disentanglement in VAEs due to high-level abstraction extraction.
method Progressive learning of independent hierarchical representations from high to low levels.
result Improved disentanglement demonstrated on two benchmark datasets using new metrics.
Proposes an extended disentanglement framework with new metrics.
problem Improving disentangled representations in representation learning.
method Connects DCI framework to identifiability, introduces new measures.
result Establishes a formal link between disentanglement and independent component analysis.
The paper defines and analyzes feature complexity in DNNs, proposing metrics for feature disentanglement and evaluation.
problem Understanding and quantifying the complexity of features learned by deep neural networks.
method Proposes a definition and disentanglement of feature complexity orders, introduces metrics for reliability and over-fitting evaluation.
result Establishes a relationship between feature complexity and DNN performance, and proposes a generic mathematical tool for network compression and knowledge distillation.
New technique learns causally disentangled representations for better generation.
problem Learning disentangled representations for accurate generation.
method Causally Disentangled Generation (CDG) approach with supervised regularization.
result CDG is necessary and sufficient for accurate disentangled generation.
New definition of disentanglement for non-independent factors of variation.
problem Current disentanglement definitions assume independent factors of variation, limiting their applicability.
method Definition based on information theory, related to Information Bottleneck Method, proposed measurement method.
result Proposed method correctly measures disentanglement with non-independent factors of variation.
Linear disentangled representations improve unsupervised action estimation.
problem Learning linear disentangled representations for unsupervised action estimation.
method Developed a method to induce irreducible representations in VAE models without labeled action sequences.
result Linear disentangled representations are a desirable property for unsupervised action estimation.
GCAE uses density estimation to achieve reliable disentanglement in latent space.
problem Disentangled learning representations suffer from reliability issues.
method GCAE uses Gaussian Channel Autoencoder with Dual Total Correlation (DTC) to avoid the curse of dimensionality.
result GCAE achieves highly competitive and reliable disentanglement scores.
New metrics assess disentanglement in generative models.
problem Measuring disentanglement in generative models is challenging.
method Introduces manifold entropic metrics based on information theory.
result Ranking of model architectures and training procedures.
Disentangled representations naturally emerge in multi-task learning.
problem Finding adaptable representations for multiple tasks.
method Empirical study of neural networks trained on automatically generated supervised tasks.
result Disentanglement naturally occurs during multi-task learning.
Disentanglement-PyTorch library facilitates disentangled representation learning.
problem Unsupervised learning of disentangled representations.
method Modular library for variational algorithms, decoupling architectures, latent space, and training algorithms.
result Achieved 3rd rank in NeurIPS 2019 Disentanglement Challenge.
GEOMANCER learns manifold factors without supervision.
problem Learning to factorize Lie group orbits from data.
method Subspace diffusion estimation of invariant subspaces.
result Unsupervised disentanglement of complex manifolds possible.
A new VAE model improves disentanglement with identifiable latent factors.
problem Learning disentangled representations without inductive biases.
method A novel VAE with a conditional prior over latent variables.
result Superior performance in disentanglement metrics.
Improved disentanglement of data factors using recursive training.
problem Current unsupervised disentanglement methods are inconsistent and fail to achieve levels of disentanglement seen in supervised approaches.
method Introduced PBT for VAEs, used UDR for heuristic scoring, and developed recursive rPU-VAE approach.
result Recursive training leads to robust disentanglement of data factors across multiple datasets.
We propose orthogonality as a necessary condition for disentangling aleatoric and epistemic uncertainty.
problem Jointly estimating aleatoric and epistemic uncertainty is problematic and non-trivial.
method We propose orthogonality as a necessary condition for disentanglement and construct UDE to measure orthogonality and consistency.
result Orthogonality and consistency are necessary and sufficient criteria for disentanglement.
DiSeNE generates interpretable node embeddings without supervision.
problem Lack of interpretability in unsupervised node embeddings.
method Disentangled representation learning with novel objective functions and metrics.
result DiSeNE produces interpretable node embeddings aligned with graph structure.
The ability to learn disentangled representations that split underlying sources of variation in high dimensional, unstructured data is important for data efficient and robust use of neural networks. While various approaches aiming towards this goal have been proposed in recent times, a commonly accepted definition and …
Proposes GM Score to evaluate GANs considering diversity, disentanglement, and discriminability.
problem Evaluation of GANs for sample quality and diversity.
method Integrates various factors including intra-class and inter-class diversity, disentanglement, and discriminability metrics.
result Demonstrates improved evaluation of GANs on MNIST dataset.
Method ranks generative models without needing latent factor supervision.
problem Challenges in selecting generative models for qualities like disentanglement.
method Ranking generative models based on training dynamics, without requiring labels for latent factors.
result Method correlates with supervised disentanglement metrics and can predict downstream performance.
Quantizes latent space to improve disentanglement in models.
problem Learning disentangled representations from unlabeled data.
method Quantizes latent space into discrete code vectors with a learnable scalar codebook and applies high weight decay regularization.
result Quantized-latent autoencoder (QLAE) outperforms prior work in disentanglement without sacrificing data reconstruction.
The paper studies how neural policies can be interpreted using decision trees.
problem Understanding how machine learning controllers make decisions in complex environments.
method The approach involves disentangled representation using decision trees to interpret neural policies.
result The paper shows that disentanglement of learned neural dynamics improves interpretability.
The key idea behind the unsupervised learning of disentangled representations is that real-world data is generated by a few explanatory factors of variation which can be recovered by unsupervised learning algorithms. In this paper, we provide a sober look at recent progress in the field and challenge some common assump…
The paper formalizes criteria for non-spurious and disentangled representations using causal methods.
problem Formalizing criteria for non-spurious and disentangled representations in representation learning.
method Causal perspective, counterfactual quantities, observable consequences of causal assertions.
result Computable metrics for assessing representation learning based on observed data.
Learning the disentangled representation of interpretable generative factors of data is one of the foundations to allow artificial intelligence to think like people. In this paper, we propose the analogical training strategy for the unsupervised disentangled representation learning in generative models. The analogy is …
Improves disentangled representation learning with multi-stage modeling.
problem Trade-off between disentanglement and reconstruction quality in autoencoders.
method Penalty-based disentanglement learning followed by detail information modeling.
result Higher reconstruction quality than state-of-the-art methods with equivalent disentanglement.
New method removes unwanted information from representations efficiently.
problem Learning representations that are uninformative about a target variable.
method Adversarial training with a novel proxy metric for mutual information, leading to an analytically computable approximation.
result Our method effectively removes unwanted information with limited time budget.
We propose a novel VAE-based deep auto-encoder model that can learn disentangled latent representations in a fully unsupervised manner, endowed with the ability to identify all meaningful sources of variation and their cardinality. Our model, dubbed Relevance-Factor-VAE, leverages the total correlation (TC) in the late…
Method evaluates disentanglement in DLVMs, including those not aligned with latent axes.
problem Evaluate disentanglement in DLVMs, especially those not aligned with latent axes.
method Proposes a statistical method to discover generative factors of a dataset.
result Empirically demonstrates the advantage of the method on two datasets.
A new method for disentangled representations without supervision.
problem Learning disentangled representations in unsupervised learning.
method Constr-DRKM, a deep kernel method with orthogonality constraints.
result Constr-DRKM performs similarly to β-VAE on disentanglement metrics.
UT module refines VAE latent space, improving disentanglement and interpretability.
problem Irregular latent distributions cause posterior collapse and misalignment in VAEs.
method UT module uses G-KDE clustering, GM modeling, and PIT to transform latent space into uniform distribution.
result UT module enhances disentanglement and interpretability of latent representations.
We propose the factorized action variational autoencoder (FAVAE), a state-of-the-art generative model for learning disentangled and interpretable representations from sequential data via the information bottleneck without supervision. The purpose of disentangled representation learning is to obtain interpretable and tr…
Improved disentanglement in VAEs using aggregated feature maps.
problem Improving disentanglement in Variational Autoencoders (VAEs).
method Regionally aggregated feature maps extracted from pre-trained CNNs on ImageNet.
result 2nd place in NeurIPS 2019 disentanglement challenge.
Improved disentanglement through learned feature aggregation.
problem Disentangling latent factors in images.
method Variational autoencoder trained on regionally aggregated feature maps from ImageNet.
result 2nd place in NeurIPS 2019 disentanglement challenge.
New framework for disentangling graph node and edge features.
problem Learning disentangled representations for attributed graphs with node and edge features.
method Proposes a novel variational objective and architecture for node and edge deconvolutions to disentangle latent factors.
result Demonstrates effectiveness of the proposed model and its extensions on synthetic and real-world datasets.
New framework learns disentangled causal representations from observed labels.
problem Learning meaningful disentangled causal representations from observed data.
method ICM-VAE framework using flow-based diffeomorphic functions and causal disentanglement prior.
result Induces highly disentangled causal factors and improves robustness.
Proposes a new method for disentangling data representations using topological analysis.
problem Learning disentangled representations for better model explainability and robustness.
method Integrates a multi-scale topological loss term into the training of deep learning models.
result Improves disentanglement scores compared to state-of-the-art methods.
A new method for disentangling action sequences improves model stability.
problem Challenges in unsupervised disentanglement learning due to incomplete theories and abstract notions.
method Introducing disentangling action sequences and a novel fractional variational autoencoder (FVAE) framework.
result FVAE improves the stability of disentanglement for action sequences.
The paper connects disentanglement to manifold charts and commutativity.
problem Discovering local charts of the data manifold for disentanglement.
method Interpreting disentanglement as local charts of the data manifold and studying commutativity.
result Commutativity is a central property in disentanglement, as shown in manifold, group theoretic, and probabilistic frameworks.
Improved VAE learns disentangled representations with less supervision.
problem Learning disentangled representations is challenging.
method Semi-supervised disentanglement learning with label replacement.
result Significant improvement in disentanglement with minimal supervision.
This work proposes a model to disentangle image factors effectively and control their manipulation.
problem Controlling disentanglement during image editing while preserving object identity.
method Encoder-decoder architecture with decorrelation regularization and soft target representations.
result The model successfully disentangles image factors and manipulates them effectively.