‘Artificial Intelligence shall avoid shortcut solutions’ says MIT
A new study by researchers at MIT finds out the problem of shortcuts in a popular machine-learning method and proposes a solution that can prevent shortcuts by forcing the model to use more data in its decision-making. By removing the simpler characteristics which the model is focusing on, the researchers force it to focus on more complex features of the data also. The researchers are trying to solve the same task in two different ways, i.e., once using those simpler features, and then using the complex features it has now learned to identify. As a result, they reduce the tendency for shortcut solutions and boost the performance of the model.
An important application of this work is to increase the effectiveness of machine learning models that are used to identify disease in medical images. Shortcut solutions could lead to false diagnoses and have dangerous implications for patients. The research will be presented at the Conference on Neural Information Processing Systems in December.