A simple sentiment analysis using a large movie review dataset. The input sentences are pre-processed using feed-forward Neural-Net Language Models and feeded to a classifier. Here, we mainly focus on the robustness of classifiers when the train output labels are randomly flipped.