About the Journal

Latent Manifold, published by Tomorrow's Research Publications (TRP), is a peer-reviewed, open-access journal covering representation learning and the geometry of learned data spaces — how neural networks compress raw data into simpler, structured internal representations, and what those representations reveal about the data itself. We publish original research and critical reviews on generative modeling, embedding theory, and the mathematical foundations of deep learning.

Scope. We welcome submissions on (but not limited to):

  • Representation learning and embedding space theory
  • Generative models (VAEs, GANs, diffusion models, normalizing flows)
  • Manifold learning and dimensionality reduction
  • Theoretical foundations of deep learning (optimization landscapes, generalization)
  • Self-supervised and contrastive learning
  • Latent space interpretability and disentanglement
  • Applications of representation learning across vision, language, and multimodal data
  • Geometric and topological methods in machine learning

Peer review. Every submission goes through double-blind peer review by at least two reviewers with relevant subject expertise, coordinated by our editorial board. We aim to give authors a timely first decision, and reviewers with a personal, financial, or institutional connection to a submitted manuscript are recused from the process.

Open access. All articles in Latent Manifold are published under a Creative Commons Attribution (CC BY 4.0) license, with no subscription or paywall for readers, and no fee charged to authors for publication.

Publication ethics. Latent Manifold follows Committee on Publication Ethics (COPE) guidelines on authorship, data integrity, plagiarism, and conflicts of interest. Submissions are screened for originality before entering peer review, and we maintain a clear process for handling misconduct allegations, including correction and retraction where warranted. Authors are expected to make code and trained models available where feasible (e.g., via GitHub with a permanent archive, or a model repository) to support reproducibility.

Who we are. Latent Manifold is edited by a board of active researchers working across machine learning theory, generative modeling, and computational mathematics. The journal is published by Tomorrow's Research Publications (TRP), an independent open-access publisher building rigorously reviewed journals in emerging and underserved research areas.