L1 Loss-Deep Learning
L1 Loss To reduce the error, which is the total of all absolute deviations between the true value and the forecast value, one uses the L1 Loss Function. Least Absolute Deviations refers to the L1 Loss function. as well as LAD. L2 Loss Function is often preferred in most situations. However, the L2 Loss Function fares poorly when there are outliers in the sample. The reason for this poor performance is that taking into account squared differences results in a considerably higher inaccuracy if the sample contains outliers. L2 Loss Function is not applicable in this situation. Use L2 Loss Function after removing the outliers if possible, however L1 Loss Function is preferred as it is unaffected by them.