Transfer learning is a very useful and important approach used in practical machine learning - especially for image classification & natural language processing. It refers to storing knowledge gained while solving a problem and using that same knowledge to solve another different but related problem. A simple example would be utilizing an image data model trained to identify cars for the purpose of identifying trucks. Machine learning guru Andrew Ng cites transfer learning (or TL) as the next driver of ML commercial success after supervised learning.
In this workshop, our expert guest speaker will show you how to leverage industry-grade image detection models to solve your computer vision problems. Discover how to extract features of your images for traditional machine learning algorithms like random forest or SVM and also howto fine-tune deep learning models like VGG16 or AlexNet. We’ll go over a particularly interesting data set for classifying birds inspired by the presenter's recent competition & show you step by step how he approached the problem and re-purposed the model. We will be using PyTorch on Google Colab - giving us the power of GPU training without any installation required.
This session will be conducted by Syafiq Kamarul Azman, research engineer at Khalifa University and a self-professed ‘machine learning geek’. Prerequisites: Advanced session - familiarity with Python, linear algebra & machine learning basics a must.
Transfer Learning: No Data? No Problem!
Transfer learning is a very useful and important approach used in practical machine learning - especially for image classification & natural language processing. It refers to storing knowledge gained while solving a problem and using that same knowledge to solve another different but related problem. A simple example would be utilizing an image data model trained to identify cars for the purpose of identifying trucks. Machine learning guru Andrew Ng cites transfer learning (or TL) as the next driver of ML commercial success after supervised learning.
In this workshop, our expert guest speaker will show you how to leverage industry-grade image detection models to solve your computer vision problems. Discover how to extract features of your images for traditional machine learning algorithms like random forest or SVM and also howto fine-tune deep learning models like VGG16 or AlexNet. We’ll go over a particularly interesting data set for classifying birds inspired by the presenter's recent competition & show you step by step how he approached the problem and re-purposed the model. We will be using PyTorch on Google Colab - giving us the power of GPU training without any installation required.
This session will be conducted by Syafiq Kamarul Azman, research engineer at Khalifa University and a self-professed ‘machine learning geek’. Prerequisites: Advanced session - familiarity with Python, linear algebra & machine learning basics a must.
Colab notebooks used in the session: