T04: Transfer Learning

 Image

Transfer learning is a machine learning technique where a model that is trained with a large dataset is applied to an unknown dataset.

Benefits: 

  • Significantly reduce the time and resources needed to develop and train a new model from scratch.
  • Transfer learning can lead to better performance when the model is applied to the new task, especially when the available data for the new task is limited. It also helps in preventing overfitting and can speed up the training process.


  • In Transfer learning we need to change the last output layer to match to output classes of new problem dataset.
  • Pre-Trained model will be freeze and only the last output layer will be trained with new dataset and ensure the already existing model will not be modified with new data.


Code: 
Cat and dog Dataset example : 
https://colab.research.google.com/github/tensorflow/examples/blob/master/courses/udacity_intro_to_tensorflow_for_deep_learning/l06c01_tensorflow_hub_and_transfer_learning.ipynb

Exercise :
exercise- flower classification
Solution - flower classification


References:
https://medium.com/towards-data-science/understanding-your-convolution-network-with-visualizations-a4883441533b


Comments

Post a Comment

Popular posts from this blog

Devops overview

Image
  Course link:  https://learn.kodekloud.com/user/dashboard CI/CD pipeline : Package dependency: Dockers: Kubernets If any container has any issue , kubernet will help to restart or resume the container operations. Terraforms Provisioning of servers if any manual change in server configuration, it will be returned to the same state again. linux interface  : link VI Editor Commands :  https://www.redhat.com/sysadmin/introduction-vi-editor user accounts:  whoami id su aparna  ssh aparna@192.168.1.2 user accounts: sudo ls /root download files :  curl URL -O wget URL -O some-file.txt os version  ls /etc/*release* Package Managers Centos uses RPM(Red Hat package  manager) software is bundled in .rpm (telnet.rpm) installing package using command - (rpm -i telnet.rpm) uninstalling package using command - (rpm -e telnet.rpm) query package using command - (rpm -q telnet.rpm) yum loads all the dependencies of a package manager Services when servers are ...
Read more

EdgeAI

Mediatek's on-Device AI solution https://neuropilot.mediatek.com/ Send a mail to  neuropilot@mediatek.com  for the tool access dimnesity900 product
Read more

GIS Dashboard for RailwayLine Safety

Image
 Requirements key features that can be extracted and visualized from RGB data collected by drones for railway lines, along with standard accepted values for a dashboard suitable for use in GIS (Geographic Information System) software: 1.       Track Alignment and Geometry: a.       Feature: Railway track centerline, curvature, and alignment. b.       Standard Values: Curvature radius, track gauge, track width, and alignment coordinates. 2.       Track Condition Assessment: a.       Feature: Track defects, cracks, and wear. Standard Values: Severity of defects (e.g., minor, moderate, severe), length and width of defects, and coordinates of defect locations. 3.       Vegetation Encroachment: Feature: Vegetation coverage along the railway corridor. Standard Values: Vegetation density (percentage coverage), distance from track centerline ...
Read more