A Review on Hidden Debts in Machine Learning Systems
Dev Kumar Chaudhary, Sandeep Kumar Srivastava, Vikas Kumar · 2018
In the present situation, Machine Learning provides a very sturdy toolkit for building applicable and useful complex prediction systems by providing new methods for creating high-show software. Unlike traditional software engineering, the collection of training examples is performed, and then the algorithms are applied. It is the hidden agenda behind artificial intelligence. Despite this fact, it is dangerous to think that this fast triumph comes for no cost. Glaring restrictions still exists in the measurements utilized and how much outcomes are imparted back to their radiating areas. We are evident of several ML-specific unpredictable factors to account for in the system design. These include algorithmic inclinations, technical arrears, data dependencies, configuration issues, the problem of authentication, granting-approval, along with faith, and human bias. We mean to motivate the continuous exchange and concentrate on coordinating the machine learning research.