Objective Cosmetic Gel Evaluation: A Deep Learning Approach Utilizing Time-Series Friction Analysis
Objective Cosmetic Gel Evaluation: A Deep Learning Approach Utilizing Time-Series Friction Analysis
The evaluation of cosmetic gels, and topical products in general, traditionally relies on subjective sensory panels. However, inherent variability in human perception, influenced by factors such as age, race, and individual sensory differences, introduces limitations to this methodology. This study presents a novel approach leveraging deep learning to objectively analyze and classify cosmetic gels based on their physical properties.
The core of this methodology involves the acquisition of time-series friction data obtained through the controlled rubbing of cosmetic gels. This raw data, reflecting the viscoelastic properties of the material, undergoes preprocessing using two distinct techniques: the Short-Time Fourier Transform (STFT) and the Continuous Wavelet Transform (CWT). These transforms extract relevant frequency components, revealing dynamic changes in the friction signals over time.
The processed data then serves as input for a deep learning model. Specifically, a ResNet-based Convolutional Neural Network (CNN) architecture was employed. The network’s performance was optimized using a learning rate scheduler, a technique designed to improve convergence and generalization. Comparative analysis was conducted using both 1D and 2D CNN models, with the STFT-processed data feeding into both types of networks, and CWT data used for a 2D CNN. The results demonstrated a superior performance of the 2D CNN model trained on STFT-processed data.
Furthermore, the robustness and reliability of the optimized STFT-based 2D CNN model were validated through k-fold cross-validation. This rigorous testing confirmed the model’s consistent performance across various data subsets. The findings suggest a promising alternative to traditional sensory panel evaluations, offering a more objective and repeatable method for assessing the user experience associated with cosmetic gels.
This deep learning approach represents a significant advancement in cosmetic product evaluation, paving the way for a more standardized and scientifically rigorous assessment of texture and feel. The objective nature of this method eliminates subjective biases, leading to more reliable and reproducible results for quality control and product development within the cosmetics industry.
Disclaimer: This content is aggregated from public sources online. Please verify information independently. If you believe your rights have been infringed, contact us for removal.