Open Access Received: 04 August 2025   |   Accepted: 29 October 2025   |   Published: 02 April 2026

MACHINE LEARNING AND AI IN POLYMER SCIENCE: A REVIEW OF PROGRESS, CHALLENGES, AND FUTURE DIRECTIONS

Ilnar N. Nurgaliev^*, Murad B. Marasulov^*, Akbarxon I. Hamzayev^*, Akmal B. Abilkasimov^**

^* Institute of Polymer Chemistry and Physics Uzbekistan Academy of Sciences, 100128, Tashkent, Uzbekistan
^** Kimyo international university in Tashkent, 156, Shota Rustaveli Street, Yakkasaray District, Tashkent, Uzbekistan, 100121

---

Abstract. Recent years have witnessed a profound transformation in polymer science driven by the rapid integration of machine learning (ML) and artificial intelligence (AI). This review provides a comprehensive and up-to-date analysis of advances in ML applications across the polymer field from 2023 to July 2025. We examine the use of supervised, unsupervised, and reinforcement learning methods for polymer design, property prediction, structural characterization, process optimization, and sustainable materials development. Special attention is given to emerging paradigms such as high-throughput screening, inverse design, multi-scale modeling, and the use of generative models—including variational autoencoders (VAEs), graph neural networks (GNNs), and transformer-based architectures. We also explore recent innovations in explainable AI (XAI), physics-informed neural networks (PINNs), and the growing role of automated experimental platforms. Key challenges—including data scarcity, model generalization, and interpretability—are discussed alongside strategies such as transfer learning, active learning, and the development of polymer-specific representations like BigSMILES. The review concludes with future directions and the outlook for AI-powered polymer discovery, highlighting the increasing role of open-access databases, multi-modal learning, and autonomous laboratories. Together, these developments mark a paradigm shift in how polymers are conceived, characterized, and optimized—ushering in a new era of intelligent materials innovation.

---

Key words. machine learning, polymer science, materials genome, property prediction, inverse design, explainable AI, physics-informed neural networks, sustainability, high-throughput screening, multiscale modeling


DOI: 10.66640/UJP-2026-5-00001

*Corresponding author: Ilnar N. Nurgaliev. Institute of Polymer Chemistry and Physics Uzbekistan Academy of Sciences, 100128, Tashkent, Uzbekistan

Citation: Ilnar N. Nurgaliev, Murad B. Marasulov, Akbarxon I. Hamzayev, Akmal B. Abilkasimov , MACHINE LEARNING AND AI IN POLYMER SCIENCE: A REVIEW OF PROGRESS, CHALLENGES, AND FUTURE DIRECTIONS. Uzbekistan Journal of Polymers, Vol. 5(1) 2026: pp.5-41. DOI: 10.66640/UJP-2026-5-00001

©2026 Uzbekistan Journal of Polymers