On 29 September, CQE hosted another edition of the CQE BreakThrough Talks, featuring Prof. Zhen Song. In his seminar, entitled “AI for Solvent (Property) Prediction”. Prof. Song presented machine learning strategies for predicting physicochemical properties of pure solvents and solvent mixtures. The talk addressed the challenges posed by sparse experimental datasets and highlighted approaches such as multi-task learning, transfer learning, pre-training and hybrid models combining artificial intelligence with thermodynamic principles. Prof. Song also introduced ai4solvents, an open web platform for high-throughput solvent property prediction and computer-aided solvent screening, supporting the rational design of greener solvents and more efficient chemical processes.
