Tensorflow for Beginners

Discover TensorFlow foundations, including tensors, operations, graphs and sessions, and common loss functions and optimizers. Learn about layers, estimators, and practical installation and running examples.

Install and run TensorFlow on mac, linux, and windows with pip installation and the GPU version, then explore simple tensor operations in Jupiter notebook and Python sessions.

Explore the machine learning lifecycle, from defining objectives to deployment and monitoring, using a toy dataset and TensorFlow. Learn to integrate TensorBoard for visualizing summaries, graphs, and model training insights.

Analyze income, experience, and IQ via a scatter matrix and heat map. Derive age from birth dates, then train a multivariate TensorFlow model to predict income.

Learn to use TensorBoard to save and view model summaries, track loss, and inspect graphs while exploring pipeline basics for automated pre-processing with TensorFlow datasets and one-shot iterators.

Name dataset components with a dictionary, explore tf record and text line data sets, and loop with one-shot iterators to read, preprocess with map and optional noise, and batch data.

Learn logistic regression as a binary classifier and overview neural network basics. Implement logistic regression by hand and with TensorFlow estimators, using a Titanic dataset example.

Convert sex to binary and build a logistic regression model with train-test split and softmax cross-entropy. Apply Adam optimizer and scale to batching with one-hot encoding via tf estimator classifier.

Explore single and multiple hidden layer neural networks in TensorFlow, learn linear and non-linear relationships, and study interactions and polynomial terms, with a preview of back propagation challenges.

Build and train single and multiple hidden layer neural networks in TensorFlow, using linear relationships, sigmoid activations, RMSE loss, and a 300-epoch, 100-batch workflow.

Explore building a multi-layer TensorFlow model with interaction terms, adding two hidden layers and varying node counts to improve loss versus a plain linear model.

Learn how convolutional neural networks use filters to form feature maps, apply max or average pooling, flatten to dense layers, and train a TensorFlow estimator on handwritten digits.

Explore building a convolutional neural network in TensorFlow for beginners, including input reshaping, conv and pooling layers, dropout, dense classifier, and training with softmax cross entropy and accuracy metrics.

Build and train a convolutional neural network in TensorFlow for beginners by loading data, defining an estimator, training, evaluating accuracy, and generating predictions with class probabilities.

Learn how to build deep neural networks with dense layers and dropout in TensorFlow, compare wide and deep architectures, and review vanishing gradients and deep learning use cases.

Explore predicting hand-written digits with a deep learning model, using dense layers, dropout, and softmax cross-entropy, and compare a dense network to a convolutional one for training and evaluation.

Build a simple object-oriented neural network framework with dense and dropout layers, a core model class, and essential methods like add, compile, train, predict, and evaluate.

Demonstrate solving neural networks with dense and dropout layers to build a dnn, then compile, train, evaluate, and predict with one-hot labels and softmax cross-entropy in TensorFlow.