# <center><i class="fa fa-edit"></i> Basic TensorFlow </center>
###### tags: `Internship`
:::info
**Goal:**
- [x] Understand basics behind TensorFlow
- [x] TensorFlow Graphs

**Resources:**
[Towards Data Science Page](https://towardsdatascience.com/understanding-lstm-and-its-quick-implementation-in-keras-for-sentiment-analysis-af410fd85b47)
[Adventures in Machine Learning](https://adventuresinmachinelearning.com/neural-networks-tutorial/#first-attempt-feed-forward)
[Machine Learning](https://hackmd.io/@Derni/HJQkjlnIP)

:::
### TensorFlow Graphs

PURPOSE: Create computational graphs to allow parallel operations and increased efficiency.

:::success

**Example**

$a = (b + c) * (c + 2)$ can also be expressed as:


As a graph:


Can perform $d = b + c$ and $e=c+2$ in parallel.
:::

TensorFlow computational graph in three-layer neural network:


- Tensors: nodes in graph.
- Multi-dimensional data arrays
- Ex: input tensor is 5000 x 64 x 1 for a 64-node input layer with 5000 training samples
- Rectified linear units: activation function in hidden layer after input layer
- Logit layer: final output layer
- Uses cross entropy as cost/loss function
- Relevant tensors flow to "Gradients" block
- Enter Stochastic Gradient Descent optimizer
- Backpropagation and gradient descent