[month] [year]

Dr. Naresh Manwani and  Mohit Chandra present paper at IEEE Transactions on Neural Networks and Learning Systems (IEEE NNLS-2019)

Dr. Naresh Manwani and his student Mohit Chandra presented a paper on Exact Passive Aggressive Algorithms for Learning to Rank Using Interval Labels at IEEE Transactions on Neural Networks and Learning Systems (IEEE NNLS-2019).  They presented exact passive-aggressive (PA) online algorithms for ordinal regression were presented. The proposed algorithms can be used even when there are interval labels instead of actual labels. The proposed algorithms solve a convex optimization problem at every trial. They found out an exact solution to those optimization problems to determine the updated parameters. Dr. Naresh Manwani and Mohit Chandra  propose a support class algorithm (SCA) that finds the active constraints using the Karush-Kuhn-Tucker (KKT) conditions of the optimization problems. These active constraints form a support set, which determines the set of thresholds that need to be updated. They derived update rules for PA, PA-I, and PA-II and showed that the proposed algorithms maintain the ordering of the thresholds after every trial. They provided the mistake bounds of the proposed algorithms in both ideal and general settings. They also show experimentally that the proposed algorithms successfully learn accurate classifiers using interval labels as well as exact labels. The proposed algorithms also do well compared to other approaches.

Link for the paper:  https://ieeexplore.ieee.org/document/8867949