Summary
The authors have introduced a new approach to controlled data generation using a variational auto-encoder. The primary objective of this method is to generate output with semantically meaningful control over specific data properties. This is done by learning a mapping function, which maps the desired property to the associated latent vectors. Authors further ensure that the influence of a change in one property (y) is independent of others during data generation by ensuring independence in the sampled latent vector (z) and the desired property (y) of the generated data. Furthermore, they propose techniques for incorporating out-of-distribution properties into the data generation process by considering different values for the properties (y) and incorporating the generated data into the overall training process.
Strengths
Strengths :
1. The concept of generating output with out-of-distribution properties has a broad range of applications and can be utilized to create previously unseen data for various downstream tasks.
2. The paper is well-written and easy to follow. It provides a thorough assessment of the limitations in prior works and demonstrates how the proposed method tackles these shortcomings.
3. The authors introduced a novel loss function to guarantee disentanglement among the desired properties.
Weaknesses
Major Comments :
1. The authors have proposed generating the latent vector (w) associated with a given property by utilizing both the sampled latent factor (z) and the desired property (y). However, the rationale for incorporating (z) in this mapping function remains unclear, particularly given that prior works[1] in the field have only used y. It is important for the authors to provide an explanation for why both z and y are considered in the mapping function and elucidate their impact on the overall results.
2. The authors should clarify the impact of choosing different values of y on the quality of the generated output, especially when the selected values for properties (y) deviate significantly from the original dataset.
3. It seems that the authors have incorporated out-of-distribution data in a manner that resembles various data augmentation techniques, like Mixup [2], aimed at improving performance. While they have demonstrated the potential of this method for generating new molecules, conducting similar experiments on datasets such as dsprites, 3D shapes, and other real-world datasets would provide valuable insights into the impact and advantages of this strategy for generating new data, as opposed to generating intermediate data points in interpolation.
4. The utilization of various constraints in the overall training of the model lacks clarity in the paper. The authors should provide more comprehensive details about these constraints and explain how they influence the overall results.
5. The authors should provide clear notational explanations for terms like "e," particularly in the context of L4 (loss), to enhance the reader's understanding.
6. The authors have provided information on how well their proposed method preserves the desired properties in the generated output. However, there is a lack of information regarding the impact on the quality of the generated images and molecules. It would be beneficial for readers to have a quantitative evaluation of these aspects, especially after incorporating out-of-distribution data during the training process.
7. Authors should clarify the prior distribution used for 'w' and 'z' in the overall optimization.
8. There has been a lot of literature discussion around the impossibility of learning disentangled representation in an unsupervised manner. The Authors have missed a seminal work in this direction [3]. This work has to be discussed and the current paper has to be positioned correctly in that context (despite the focus of this paper is not a fully unsupervised learning).
9. There has been work on disentangled learning beyond datasets such as Dsprites. For instance, real-world datasets, that have attributs can be considered to test the proposed idea. I recommend the Authors to consider such experiments, especially given the cliam regarding the "precisely controlling the properties of generated data".
10. There is a very closely related work that claims to do a similar type of stuff detailed in [4].
References:
[1] Guo, Xiaojie, Yuanqi Du, and Liang Zhao. "Property controllable variational autoencoder via invertible mutual dependence." International Conference on Learning Representations. 2020.
[2] Taghanaki, Saeid Asgari, et al. "Jigsaw-vae: Towards balancing features in variational autoencoders." arXiv preprint arXiv:2005.05496 (2020).
[3] Locatello F, Bauer S, Lucic M, Raetsch G, Gelly S, Schölkopf B, Bachem O. Challenging common assumptions in the unsupervised learning of disentangled representations. In international conference on machine learning 2019 May 24 (pp. 4114-4124). PMLR.
[4] Mondal, Arnab Kumar, Ajay Sailopal, Parag Singla, and Prathosh Ap. "SSDMM-VAE: variational multi-modal disentangled representation learning." Applied Intelligence 53, no. 7 (2023): 8467-8481.
Confidence
4: You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work.