I'm kinda new to machine learning and wanted to know if we could use multiple machine learning algorithms, for example, SVM and backpropagation together to solve a particular problem.
You can train multiple machine learning models with same data and based on accuracy and confusion_matrix response you can decide which one to use.
In theory, you don't need to combine/merge two different ML Models (you can enhance your pre-processing) but if you still want to use different models there are two approaches:
- Ensemble (available with three categories - a) Bagging b) Boosting c) Stacking
- Hybrid (This approach allow users to create own models (or use existing) and combine them for better prediction)
Note: You need to be careful with individual algorithm response before combining them together :)
In a classification/regression task you can use back propagation and SVM:
- Backpropagation: use a neural network as feature extractor
- SVM: use it to perform classification/regression with the features extracted with the neural network
In deed, in neural networks back-propagation and other well-known machine learning techniques are used together. For example, when a sigmoid layer is used as the classification layer for a binary classification neural net, a logistic regression is performed and optimized through back-propagation