search

Cimagroup / Experiments-Representative-datasets

GNU Affero General Public License v3.0
3 stars 0 forks source link

May you clarify how different your paper is from this paper: Investigating the Working of Text Classifiers #1

Closed Sandy4321 closed 5 years ago

Sandy4321 commented 5 years ago

https://github.com/DevSinghSachan/investigating-text-classifiers https://arxiv.org/abs/1801.06261 Investigating the Working of Text Classifiers May you clarify how different your paper is from this paper: Investigating the Working of Text Classifiers

EduPH commented 5 years ago

Hi!, I would need to read it carefully but my first impression is that they focus on natural language processing datasets. Using then a methodology really intrinsic to the type of data: lexicons.... In our case, we are dealing with any n-dimensional dataset, and, using a nearness condition, reduce the size of data, i.e., the number of points. Then, we propose as a tool to measure the similarity of two different datasets both Hausdorff distance and Bottleneck distance.

Besides, it seems that in the paper you provide, they are trying to discriminate if MLP can really generalize datasets of NLP or not.

Thank you for your interest in our paper. You can find it here.

Greetings

Sandy4321 commented 5 years ago

great thanks you explained very well do you have this kind of clear explanation in written for example presentation slides: less scientific but more practicable?

EduPH commented 5 years ago

Not yet. However, if I do some slides, I would add them to this repository.