Optimization of Material Selection for Whitening Cream: Artificial Neural Networks and Genetic Algorithm Approach

摘要

The materials in the skin care products have many types such as the whitening agent in the whitening cream. In general, researchers have conducted experiments to determine the effectiveness of such substances in the form of a single material in vitro. In practice, performance testing of material has conducted in Vivo that it is more than acceptable. In addition, combination of the materials (synergistic) makes higher performance or jump higher in a non-linear manner. The amount of these materials should be used in the right proportions. However, to test the performance of materials in the real trial each time, it has a very high cost due to volunteer, tracking, and duration of the experiment. We were unable to test the performance in several formulations. For this reason, the research objective was to develop algorithms to solve the problem of restrictions on such experiments in order to reduce time and cost. With a wide range of experimental design, we used techniques to predict the outcome in early stage, then applied those results to design experiments and collect data. In this work, the effects of artificial neural networks (ANNs) to predict the effectiveness of the materials in whitening cream products were used incorporate with Genetic Algorithm (GA). In this article, we experimented with simulated data from the experts as close to real data. To test the performance of the algorithm was developed and will extend the experimental results in the future. The algorithm of ANNs developed a multi-layer feed forward structure (4-13-1) with the lowest MSE is 6.00895e~(-4) and largest R~2 is 0.979164. The best materials formulation optimized by GA were Arbutin=3%, Aloesin=0.658%, Niacinamide=0.007%, and Oxyresveratrol=0.993% that conduct lowest of 0.0823817 melanin value. Therefore, the algorithm developed in this study can be applied to develop the reality experiment in the future.