The machine learning (ML) technique, and more specifically, a Convolutional Neural Network (CNN), was utilized as a method to anticipate tool wear. Milling is used as an example to demonstrate experimentally how the proposed methodology should be implemented. Experiments are carried out via dry machining techniques, which involve a non-coated ball end mill and a work-piece made of stainless steel. In-situ analysis of the amount of wear on the flanks is performed with the use of a digital microscope. The machine learning model's predictions are founded on an experience database that stores all of the data from the experiments that came before it. The in-process tool wear prediction system that was proposed will, at some point in the future, be supplemented by an adaptive control (AC) system. This AC system will communicate continuously with the ML model in order to seek out the optimal adjustment of feed rate and spindle speed that allows for the optimization of flank wear and the extension of tool life. The decisions made by the AC model are based on the forecast that was supplied by the ML model as well as the information feedback that was provided by the force sensor. The force sensor captures the change in the cutting forces as a function of the advancement of the flank wear. Only the machine learning model component for estimating tool wear based on CNNs has been demonstrated in this body of work. The methodology that was suggested has demonstrated an accuracy of approximately 90%. Additional experiments will be carried out to validate the repeatability of the findings, and the measuring range will be expanded in order to enhance the precision of the existing measurement system

Seep Belief Network (DBN), multi-task learning, tool wear condition, surface quality prediction