In any learning framework, an expert knowledge always plays a crucial role.\nBut, in the field of machine learning, the knowledge offered by an expert is\nrarely used. Moreover, machine learning algorithms (SVM based) generally use\nhinge loss function which is sensitive towards the noise. Thus, in order to get\nthe advantage from an expert knowledge and to reduce the sensitivity towards\nthe noise, in this paper, we propose privileged information based Twin Pinball\nSupport Vector Machine classifier (Pin-TWSVMPI) where expert's knowledge is in\nthe form of privileged information. The proposed Pin-TWSVMPI incorporates\nprivileged information by using correcting function so as to obtain two\nnonparallel decision hyperplanes. Further, in order to make computations more\nefficient and fast, we use Sequential Minimal Optimization (SMO) technique for\nobtaining the classifier and have also shown its application for Pedestrian\ndetection and Handwritten digit recognition. Further, for UCI datasets, we\nfirst implement a procedure which extracts privileged information from the\nfeatures of the dataset which are then further utilized by Pin-TWSVMPI that\nleads to enhancement in classification accuracy with comparatively lesser\ncomputational time.\n