Progressive Learning Via Rearrangement of Noisy Labels
Adlene Ebenezer P1, Shantanu Fartyal2, Manish Prakash3, Osama Habib4, Aditya Siddharth5

1Adlene Ebenezer P Assistant Professor, Department of Computer Science and Engineering, SRM IST, Ramapuram (Tamil Nadu), India.
2Shantanu Fartyal, Department of Computer Science and Engineering, SRM IST, Ramapuram (Tamil Nadu), India.
3Manish Prakash, Department of Computer Science and Engineering, SRM IST, Ramapuram (Tamil Nadu), India.
4Osama Habib, Department of Computer Science and Engineering, SRM IST, Ramapuram (Tamil Nadu), India.
5Aditya Siddharth, Department of Computer Science and Engineering, SRM IST, Ramapuram (Tamil Nadu), India.
Manuscript received on 01 May 2019 | Revised Manuscript received on 15 May 2019 | Manuscript published on 30 May 2019 | PP: 1043-1046 | Volume-8 Issue-7, May 2019 | Retrieval Number: G5625058719/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: In the recent years the innovation in machine learning has scaled up to a whole another level. Large scale learning problems need a vast variety of labels which can be collected at a low cost. Crowdsourced data offers a really low cost but it comes with a lot of noise, this means that the data collected cannot be trusted and hence can degrade the performance. Among the noisy labels, some labels can be really important. To tackle the difficulty of noisy labels and degraded performance, we offer to propose and actualize, a framework including POSTAL (Progressive Stochastic Learning of Noisy Labels), a new innovation for rearrangement of labels. Our framework gives a double arrangement. One, it sorts all the labels from the reliable one to noisy one. Two, progressively feed the data into the machine to learn.
Keyword: POSTAL, Noisy Labels, Progressive Learning
Scope of the Article: Learning Software Design Engineering