DeepDIVE: Optimizing Input-Constrained Distributions for Composite DNA Storage via Multinomial Channel

2025-01-25

AI Summary

This work addresses the polynomial channel capacity optimization problem for DNA storage under constrained input support sets. We propose a joint optimization framework integrating a variational autoencoder (VAE) with the multidimensional dynamic allocation Blahut–Arimoto (M-DAB) algorithm. Unlike conventional fixed-grid discretization methods, our approach is the first to embed the VAE directly into the M-DAB iterative procedure, enabling end-to-end joint optimization of input symbol positions and probability weights. Leveraging an alternating optimization strategy under composite DNA channel models, the framework achieves significantly improved capacity approximation accuracy: it attains 98.5% of the theoretical capacity even when the support size is reduced by 40%. This effectively overcomes the fundamental discretization bottleneck in capacity computation. The method establishes a scalable, high-precision paradigm for channel capacity optimization—critical for designing high-density, low-redundancy DNA storage systems.

Technology Category

• Search and Optimization: Mixed Discrete/Continuous Search
• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Learning & Optimization for CV

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

We address the challenge of optimizing the capacity-achieving input distribution for a multinomial channel under the constraint of limited input support size, which is a crucial aspect in the design of DNA storage systems. We propose an algorithm that further elaborates the Multidimensional Dynamic Assignment Blahut-Arimoto (M-DAB) algorithm. Our proposed algorithm integrates variational autoencoder for determining the optimal locations of input distribution, into the alternating optimization of the input distribution locations and weights.

Problem

• Polynomial Channel Optimization
• Limited Input Range
• DNA Data Storage

Innovation

• DeepDIVE algorithm
• polynomial channel input distribution optimization
• variational autoencoder