TY - GEN
T1 - Integrating Machine Learning with Limited Datasets to Optimize Azo Dye Synthesis
T2 - 5th International Conference on Data Engineering and Communication Technology, ICDECT 2024
AU - du Plessis, Claudius
AU - Lebea, Khutso
AU - Meijboom, Reinout
AU - Leung, Wai Sze
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
PY - 2025
Y1 - 2025
N2 - The synthesis of azo dyes, renowned for their vivid colors and stability, is essential in industries such as textiles, cosmetics, and food colorants. Traditional synthesis methods, involving diazotization and azo coupling reactions, are often complex, resource-intensive, and environmentally taxing due to the use of toxic chemicals and the generation of hazardous waste. As the demand for sustainable and efficient manufacturing processes grows, there is a pressing need to innovate and optimize dye synthesis techniques. This paper investigates the integration of machine learning (ML) techniques in azo dye synthesis, focusing on supervised learning, deep learning, and strategies for small datasets. By leveraging ML algorithms, researchers can predict synthesis outcomes from limited experimental data, facilitating the search for specific dyes and the optimization of known reactions. This capability is particularly beneficial in azo dye synthesis, where experimental data can be scarce and diverse. Through a detailed analysis of recent advancements and case studies, this research demonstrates how ML can enhance the search for specific dyes and optimize existing processes using minimal datasets. These findings highlight the transformative potential of machine learning in chemical synthesis, setting a new paradigm for the development of azo dyes in the modern era.
AB - The synthesis of azo dyes, renowned for their vivid colors and stability, is essential in industries such as textiles, cosmetics, and food colorants. Traditional synthesis methods, involving diazotization and azo coupling reactions, are often complex, resource-intensive, and environmentally taxing due to the use of toxic chemicals and the generation of hazardous waste. As the demand for sustainable and efficient manufacturing processes grows, there is a pressing need to innovate and optimize dye synthesis techniques. This paper investigates the integration of machine learning (ML) techniques in azo dye synthesis, focusing on supervised learning, deep learning, and strategies for small datasets. By leveraging ML algorithms, researchers can predict synthesis outcomes from limited experimental data, facilitating the search for specific dyes and the optimization of known reactions. This capability is particularly beneficial in azo dye synthesis, where experimental data can be scarce and diverse. Through a detailed analysis of recent advancements and case studies, this research demonstrates how ML can enhance the search for specific dyes and optimize existing processes using minimal datasets. These findings highlight the transformative potential of machine learning in chemical synthesis, setting a new paradigm for the development of azo dyes in the modern era.
KW - Data-driven synthesis
KW - Hyperparameter optimization
KW - Machine learning
KW - Small datasets
KW - Supervised learning
KW - Synthesis of azo dyes
UR - https://www.scopus.com/pages/publications/105015712611
U2 - 10.1007/978-981-96-5214-3_13
DO - 10.1007/978-981-96-5214-3_13
M3 - Conference contribution
AN - SCOPUS:105015712611
SN - 9789819652136
T3 - Lecture Notes in Networks and Systems
SP - 153
EP - 165
BT - Innovations in Data Engineering
A2 - Bhateja, Vikrant
A2 - Rana, Muhammad Ehsan
A2 - Tripathy, Hrudaya Kumar
A2 - Senkerik, Roman
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 28 September 2024 through 29 September 2024
ER -
