# Comprehensive Review On Twin Support Vector Machines

**Authors:** M. Tanveer, T. Rajani, R. Rastogi, Y. H. Shao, M. A. Ganaie  
**arXiv:** [2105.00336](https://arxiv.org/abs/2105.00336)  
**Date:** 2021-05-01  
**Summary page:** https://www.summarizepaper.com/en/paper/2105.00336v3/

## Key points

- TWSVM and TSVR are emerging machine learning techniques for classification and regression challenges
- TWSVM aims to identify two nonparallel hyperplanes for effective classification
- TWSVM solves two small-sized QPPs instead of a single large-sized QPP like traditional SVM
- TSVR is formulated based on the principles of TWSVM and involves solving two SVM-like problems
- Limited literature comparing different variants of TSVR exists
- The review provides a rigorous analysis of recent research in both TWSVM and TSVR, highlighting their limitations and advantages
- The paper introduces the basic theory of support vector machines before explaining TWSVM and its improvements and applications
- The concept of TSVR is discussed along with its enhancements
- Detailed insights into the strengths and weaknesses of these techniques are provided throughout the review
- Future research directions and development prospects for TWSVM and TSVR are suggested in the conclusion

## Summary

The paper titled "Comprehensive Review On Twin Support Vector Machines" provides a thorough analysis of twin support vector machine (TWSVM) and twin support vector regression (TSVR), which are emerging machine learning techniques for classification and regression challenges. TWSVM aims to identify two nonparallel hyperplanes that can effectively classify data points into their respective classes. Unlike traditional support vector machines (SVM), TWSVM solves two small-sized quadratic programming problems (QPPs) instead of a single large-sized QPP. Similarly, TSVR is formulated based on the principles of TWSVM and involves solving two SVM-like problems. These techniques have shown promising results in various applications, but there is limited literature comparing different variants of TSVR. Therefore, this review aims to fill this gap by providing a rigorous analysis of recent research in both TWSVM and TSVR, highlighting their limitations and advantages. The review begins by introducing the basic theory of support vector machines, followed by an explanation of TWSVM and its various improvements and applications. The authors then delve into the concept of TSVR and discuss its enhancements. Throughout the review, the authors provide detailed insights into the strengths and weaknesses of these techniques. In conclusion, the paper suggests future research directions and development prospects for TWSVM and TSVR. Overall, this comprehensive review serves as a valuable resource for researchers interested in understanding and utilizing these emerging machine learning techniques for classification and regression tasks.