Skip to content

adb1997/Neural-Networks

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

---HAND SIGNS---

The code is for a plain neural network built on Tensorflow to classify handsigns. Architecture of the model is as follows :

CONV2D -> RELU -> MAXPOOL -> CONV2D -> RELU -> MAXPOOL -> FLATTEN -> FULLYCONNECTED

Cost function is minimised using Softmax Cross Entropy with logits (https://www.tensorflow.org/api_docs/python/tf/nn/softmax_cross_entropy_with_logits) Accuracy on training set : 94.3 %. Can be improved by increasing number of epochs. (Hardware constraints)

---Residual Networks Hand Signs---

The same hand signs dataset is trained on a deep 50 layer neural network.

Identity block: CONV2D -> BATCHNORM -> RELU -> CONV2D -> BATCHNORM + RELU where another CONV2D -> BATCHNORM block skips the first two and gets added at the +

Full Network: CONV2D -> BATCHNORM -> RELU -> MAXPOOL -> CONVBLOCK -> IDBLOCK2 -> CONVBLOCK -> IDBLOCK3 -> CONVBLOCK -> IDBLOCK5 -> CONVBLOCK -> IDBLOCK2 -> AVGPOOL -> TOPLAYER

--- MNIST---

Some neural network classifiers for the MNIST and Fashion MNIST datasets. Experimented with regular neural network in Tensorflow as well as a Convolutional Neural Network to explore accuracy improvement.

---Character Level Text Generation---

Generating new Dinosaur names from a list of existing names using a character level language model. Gradient clipping used inside the optimization function to prevent exploding gradients. Sampling carries out generation of new words.

About

Dabbling in Deep Learning

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages