A set of examples using tensorflow for machine learning classification
Dataset originally taken from here. Full dataset available here.
To run, cd
into india_foot_height
and run python india_foot_height.py
. It'll print out the value of the cost function on each iteration of training and in the end plot the fit function on the dataset.
Description of the dataset available here, and the actual dataset is available here.
To run, cd
into iris-flower
and run python iris-flower.py
. It'll split the total dataset into training and test sets, train a neural network and then plot the value cost function over time.
A convolutional neural network for image classification, using a subset of the COIL-20 image dataset - three classes to be more precise, rubber-duck, mug, pigbank. It uses Google's pretrained Inception V3 model. To run this, you'll need:
- cd into
coil-20
mkdir images/inception-images
- this is where the pre-processed images for re-training will be kept./preprocess.sh
- this will pre-process all images under images/training, to a format that the tensorflow inception model accepts./train.sh PATH_TO_INCEPTION_V3_MODEL
- you'll need to download the Inception V3 model and use the path you saved in place ofPATH_TO_INCEPTION_V3_MODEL
. You can tune the training parameters (e.g., max iterations) by modifying thetrain.sh
script- After training, run
./eval
to see the level of accuracy you achieved