Similarity Detection using Siamese Networks
- 0 Collaborators
Siamese networks are a special type of neural network architecture. Instead of a model learning to classify its inputs, the neural networks learns to differentiate between two inputs. It learns the similarity between them. ...learn more
Project status: Published/In Market
Groups
Student Developers for AI,
DeepLearning
Intel Technologies
AI DevCloud / Xeon,
MKL
Overview / Usage
Standard Classification vs. One Shot Classification
Standard classification is what nearly all classification models use. The input is fed into a series of layers, and in the end , the class probabilities are output. If you want to predict dogs from cats, you train the model on similar(but not same) dogs/cats pictures that you would expect during prediction time. Naturally, this requires that you have a dataset that is similar to what you would expect once you use the model for prediction.
Methodology / Approach
One Shot Classification models, on the other hand, requires that you have just one training example of each class you want to predict on. The model is still trained on several instances, but they only have to be in the similar domain as your training example.
Technologies Used
intel optimized numpy, mkl, intel devcloud