Need Suggestion for a Project

AI Discussions
1

I have recently done “Supervised Machine Learning: Regression and Classification course”. I took the course because i am currently doing a project on classification. In this course, we learn about binary classification. But in my project i need to develop a multi-classifier to classify four diseases from signal data. I have extracted some features but i am not understanding how to fit my model to those features. I would be grateful if anyone could provide me with guidelines or suggest further courses that would benefit my project.

4 Likes
2

Hi @paromita_kundu

Welcome to the community.

Here some resources regards multi-class modeling:

  1. Multi-Class Neural Networks | Machine Learning - Google Developers: This course covers multi-class neural networks and provides an understanding of multi-class classification problems.
  2. Multi-Class Classification Tutorial with the Keras Deep Learning Library: This tutorial shows how to use Keras to develop and evaluate neural network models for multi-class classification problems.

Here some papers regards multi-class disease classify:
https://scholar.google.com.br/scholar?q=disease+multiclass+modeling&hl=pt-BR&as_sdt=0&as_vis=1&oi=scholart

I hope this help

best regards

2 Likes
3

You could start really simple and efficiently by using support vector machine for multiple classes:

2 Likes
4

Thank you so much @elirod for giving the link of resources and papers. I truly appreciate it!

1 Like
5

Thanks @Nydia . I will go through your suggested technique. <3

2 Likes
6

You welcome. :heart:

1 Like
7

What course did you attend specifically? Was it a course from deeplearning.ai?

Almost any respectable course in classification is going to include an example where you have multiple output labels - the most common example is recognizing images of single handwritten digits (zero through nine).

In your case, all you need to do is have an output layer with four units.

Or you can use a linear output layer and set “from_logits = True”, and TensorFlow will handle the multiple classes automatically (if your course included TensorFlow, that is).

8

Yes @TMosh , It was from deeplearning.ai.("Course 1:Supervised Machine Learning: Regression and Classification course”).They mentioned about multiple output labels in one video. But there was not such example as long as I remember. It also does not include TensorFlow so It’s difficult for me to understand your direction about linear output layer. If you have any further suggestions, please let me know.

9

I recommend you continue with the Machine Learning Specialization. The information you need is in Course 2 (“Advanced Learning Algorithms”) in Week 2 (Multiclass Classification).

2 Likes
10

I will definitely continue it. Thanks for all your recommendation @TMosh .

1 Like