Deep Learning Intoductions along with Examples.pptx

Deep Learning

1. Shivam Patil T.E(CSE-AI) 27 *Transfer Learning*

2. Introduction Transfer learning, used in machine learning, is the reuse of a pre-trained model on a new problem. In transfer learning, a machine exploits the knowledge gained from a previous task to improve generalization about another. Transfer learning (TL) is a technique in machine learning (ML) in which knowledge learned from a task is re-used in order to boost performance on a related task. For example, for image classification, knowledge gained while learning to recognize cars could be applied when trying to recognize trucks.

3. *Need for transfer learning in deep learning * Scarcity of Data : In many real-world scenarios, collecting labeled data for training deep learning models can be expensive or time-consuming. Transfer learning allows us to utilize pre-trained models trained on large datasets, reducing the need for extensive labeled data for the target task. Computational Resources : Training deep learning models from scratch can be computationally intensive, requiring significant time and resources. Transfer learning allows us to start with pre-trained models, which have already learned useful features from large datasets, saving computational resources and time.

4. Computational Resources : Training deep learning models from scratch can be computationally intensive, requiring significant time and resources. Transfer learning allows us to start with pre-trained models, which have already learned useful features from large datasets, saving computational resources and time. Domain Adaptation : Often, the distribution of data in the target task may differ from that of the source task. Transfer learning helps adapt the learned features from the source domain to the target domain, improving the model's performance on the target task. Feature Learning : Deep learning models learn hierarchical representations of data, where lower layers capture low-level features (e.g., edges, textures) and higher layers capture more abstract features. Transfer learning allows us to transfer these learned features, which are often generic across tasks, to new tasks, enabling faster convergence and better performance.

5. * Feature extraction using transfer learning* Feature extraction transfer learning is when you take the underlying patterns (also called weights) a pretrained model has learned and adjust its outputs to be more suited to your problem.

6. !!! THANK YOU !!!