I want to use a neural network classifier to separate patients and healthy persons by some parameters. and then test validation of the classifier. Is svm a proper tool? And how can I test the validation? Which Syntaxs should I use?
First, SVM and neural network are essentially different classifiers. You should use both SVM, NN and a lot of other algorithms (Naive Bayes, Logistics Regression, Decision Trees, etc.). There are a lot of articles, described of using of these algorithms for early diagnostics of Heart, Liver, IVF, Prostate and other diseases. See, for example, :
1. Animesh Hazra, Subrata Kumar Mandal, Amit Gupta, Arkomita Mukherjee and Asmita Mukherjee. Heart Disease Diagnosis and Prediction Using Machine Learning and Data Mining Techniques: A Review. Advances in Computational Sciences and Technology. Volume 10, Number 7 (2017) pp. 2137-2159
2. Article A Review on Heart Disease Prediction using Machine Learning ...
Sergey Porotsky Thank you for your advice. I'll review the article.
I accept previous answers by researchers.Additionally, why do not you use Neutral network tool for pattern recognition in MATLAB?it may gives what you are looking for..Also,I think SVM works too..
This link contains SVM classifier file exchange but only for 2 classes
https://in.mathworks.com/matlabcentral/fileexchange/63158-support-vector-machine
Aparna Sathya Murthy Thank you for replying. Yes, my data is labeled. I don't want to extract features. Just want the classifier to learn the relation between my features and labels and can predict a new sample. And I want to use MATLAB
Bazhdar Noureddin Mohammed Thank you for your guide. I have 2 classes too. Maybe I'd better try neural network tool as you suggested.
Mahshid Yousefian Based on your discussion above, NN is better as a predictor or classifier.
I prefer to write Matlab code by yourself.
You can read one of many articles and books in the field like the following:
Mallick, Pradeep Kumar; Balas, Valentina Emilia; Bhoi, Akash Kumar; Zobaa, Ahmed F.[Advances in Intelligent Systems and Computing] Cognitive Informatics and Soft Computing Volume 768 (Proceeding of CISC 2017) || Functional Link Artificial Neural Network (FLANN) Based Design of a Conditional Branch PredictorSpringer-Verlag 2019.
Thanks and regards
Additional comments:
1. SVM excellent handle non-linear relations by means of different kernels using (RBF, Poly, etc.)
2. For large Data Sets and a lot of parameters (as images, texts, speech, etc.) using of NN (rather, Deep Learning) is essentially better. But for regular tasks, as early diagnostics of diseases (10-20 features, less than 10,000 patients) SVM often gets us better results - see, please articles of different methods comparison.
3. SVM provide convex goal function (quadratic programming) and NN should use global optimization due to non-convex goal function.
My advice - to start from simple method (as Naïve Bayes, Logistic Regression, SVM, etc.) and if it will be necessary, to use NN and Deep Learning.
Asslamulakam
I have attached a file which has a command how to use the neural net work classifier in Matlab
Have nice time
Regards
zahid hasan
- Badges
- Science topic