Complete AWS Machine Learning for Beginners and Masters

Explore how to store and fetch data in Amazon S3 using Python in Jupyter notebooks and SageMaker, with hands-on manual uploads and programmatic downloads.

Explore Amazon S3, a simple storage service that uses buckets and files to store data for apps, backups, and SageMaker machine learning, with easy upload and retrieval.

Explore storing data in S3 with SageMaker, both manually and programmatically, by creating a bucket with permissions and uploading a text file and an image via the S3 uploader.

Download a large dataset from a URL, convert it into a dense tensor, and store it in an S3 bucket using boto3 in SageMaker for bulk data uploads.

Learn to fetch data from S3 into SageMaker Studio, download text and image files using boto3, and read them as text or binary data for model input.

Explore storing and fetching data with S3 and SageMaker using three buckets, boto3, and manual uploads, then apply these methods to real machine learning projects.

Explore how to build and deploy a binary classifier in SageMaker using MNIST digits with the linear learner, covering data preparation, training, and endpoint deployment.

Explore the amnesty dataset, a 28 by 28 grayscale collection of handwritten digits zero through nine. Download, format, and host it in sage maker with s3 to train and validate.

Learn to set up a SageMaker notebook, download and unzip the Manistee dataset, create train/validation/test splits, and prep data for S3 deployment.

Format a dataset for a model by converting image inputs to float tensors, labeling zeros vs non-zeros, and uploading training data to an s3 bucket for SageMaker.

Explore how a linear learner uses multiple parallel linear models with different hyperparameters to select the best performing classifier or regressor in supervised learning.

Build and train a linear learner model with a SageMaker estimator, configuring hyperparameters, 784 feature dimensions, and a binary classifier, then train on S3 data using the fit function.

Deploy and host the trained model, then validate its performance with test data. Create a confusion matrix to evaluate predictions and access via a Python SDK.

Delete the endpoint to prevent usage and stop billing after finishing a model; SageMaker saves the model and data in S3, and you can redeploy later.

Build, deploy, and validate a binary zero-vs-nonzero digit classifier with a linear learner in SageMaker using the Manistee digits dataset. Explore data preparation and endpoints.

Explore how to use the SageMaker debugger to monitor training, apply pre-built debugging rules, and inspect model structure and tensor values to diagnose training issues.

Learn how the SageMaker debugger monitors training, detects issues like vanishing gradients and loss not decreasing, and pinpoints problematic tensors with rules, in parallel across metrics.

Create a new project in the stage studio, import prebuilt models into the debugger, and configure entry-point scripts and a Python notebook to begin training on SageMaker.

Build an estimator using a tensor flow container, set rules to detect vanishing gradient and loss not decreasing, and train with the NIST image dataset.

Print the training job summary and logs, inspect rule evaluations for vanishing gradient and loss not decreasing, and identify issues using SageMaker and S3 stored artifacts.

Examine tensor performance during training by tracking losses across iterations, create trials from debug artifacts, and plot loss curves with matplotlib to assess model progress.

Expose students to using stage maker debugger to track model performance during training, apply predefined rules on a TensorFlow-based estimator, and visualize tensor values and loss over iterations.

Learn to import Saikat Learn data into the Sage Maika ecosystem, set up jupyter notebook projects, and build a model to classify iris species.

Explore how scikit-learn integrates with SageMaker to import data, build and train an estimator, and test a model that classifies iris species.

Explore the iris dataset with four features in centimeters to classify 150 data points (50 per species) into three species, using labels 1-3 in a csv file.

Set up a new SageMaker project and notebook for iris speciation, import the script, and configure S3 data and model paths to train a decision tree classifier.

Download and format the iris dataset, join inputs and labels into a single csv, save it in a local data folder, and upload to S3 for SageMaker use.

Create and train an Escalon container in SageMaker that points to iris CSFI data in S3 and checkpoints, setting the entry point, training instance, and max_leaf_nodes.

Deploy the model, create a predictor, and test predictions against actual test data to evaluate accuracy, then delete endpoints to avoid extra usage.

Learn to build, train, and deploy a simple iris species classifier with SageMaker and Saikat Learn, using a Python script, containers, and a notebook workflow.

This course introduces using the boost library in AWS SageMaker to build a multiclass classifier on the Manistee data set. It identifies handwritten digits.

Discover XGBoost, a distributed gradient boosting library that speeds training with parallel tree boosting. It scales to millions or billions of data points and runs across languages and platforms.

Set up a new AWS machine learning project in SageMaker, create three S3 buckets, and configure the notebook workflow for data retrieval and path setup.

Download and format the Manistee dataset for training, then upload the converted train, validation, and test sets to S3, using binary buffers and latin-1 encoding for reliable parsing.

Train XGBoost models by creating a boost container, configuring common and specific training parameters, and running single-machine and distributed jobs in parallel on S3 data.

Deploy and host trained models by importing the model, configuring endpoints, and creating hosted endpoints with production variants; monitor statuses and validate deployment through endpoint descriptions.

Test and validate the model by predicting a single value, batch predictions, using a confusion matrix to visualize performance across digits zero through nine, and delete endpoints to conserve resources.

Review the code to extract data, split into train, validation, and test sets, and format for S3 uploads. Train and host models, run predictions, and evaluate with a confusion matrix.

Apply xgboost in sage maker to train a multiclass handwritten digit classifier using the mnist dataset, then host and validate the model while exploring other classification tasks.

Explore SageMaker flow container basics. Use a tenzer flow container with prewritten scripts to tune models, set hyperparameters, and manage data.

Explore the TensorFlow library, its computational graphs, tensors and neural networks, and how SageMaker runs pre-written models within a flexible, layer-by-layer workflow.

Explore the mnist dataset of 28x28 grayscale handwritten digits (0–9) with 60,000 training, 10,000 testing, and 10,000 validation images, all pre-labeled and mutually exclusive for reliable image classification.

Set up a new AWS SageMaker project, create and attach an S3 bucket, and import the model and utilities scripts to enable tensor flow tuning with hyperparameter optimization.

Examine the mnist utils and model scripts, convert images and labels to tfrecord files, and build a convolutional neural network with pooling, dropout, and softmax for training, evaluation, and prediction.

Download the MDT dataset from the TensorFlow library, extract and format it into train, validation, and test TF records, and upload the data to an S3 bucket under a prefix.

Create and configure the model containers in SageMaker, set a continuous learning rate range, define loss-based objective metrics, and launch a hyperparameter tuner to run nine trials at a time.

Launch and monitor a hyperparameter tuning job, print its status, and observe progress on the SageMaker dashboard as three concurrent runs complete.

Learn Tenzer flow tuning by building a TensorFlow model in a tense flow container, using an amnesty script and a tuner to train, tune hyperparameters, and explore Sage Maker deployment.

Build and deploy an image classification model using SageMaker, starting from Python in a Jupiter notebook style environment to train, host on S3, and create endpoints for inference.

Set up a new AWS machine learning project in SageMaker Studio, create or reference an S3 bucket, and import the image classifier notebook.

Download the Kaltech 256 dataset locally, extract training and validation sets, and upload them to S3 as training artifacts for model development.

Learn to create a training job in AWS machine learning, configuring a model container with 18 layers, 257 classes, and 64 mini-batches, using S3 data paths, and monitoring status.

Create an inference model in AWS SageMaker from trained artifacts in S3, configure a hosting image and a batch transform job, and prepare input data for image classification inferences.

Deploy and host your trained model by creating an endpoint configuration and endpoints, then monitor their status until in service, enabling real-time inference.

Perform real time inference on a hosted AWS ML model by using an endpoint and runtime to test a bathtub image, display predictions and probabilities, and manage endpoints.

Learn to build and deploy an image classification model with SageMaker Studio on Kaltech 256 dataset, train and create an inference model, deploy endpoints, and run inferences.

Build and train a movie-genre predictor from plot data using AWS SageMaker, Python, and Jupiter notebooks, while processing genre, plot, and vocabulary datasets.

Set up a complete aws machine learning project by creating a new studio in SageMaker, configuring S3 buckets, and following a cell-by-cell notebook workflow.

Fetch and explore the movies datasets by downloading and extracting genres and plots, build data frames, apply one-hot encoding, and prepare English-focused vocabulary for modeling.

Fetch and refine the vocabulary dataset for plot analysis by tokenizing text, removing stop words, and building a comprehensive vocab to feed a sentiment analysis model.

Create and train a model by configuring parameters and setting up a container. Launch training jobs and manage training, validation, and test data in S3 with glove embeddings and vocab.

Evaluate model performance by building an evaluation function, reading correct answers and predictions, and computing accuracy, precision, recall, and F1 across genres using a 0.6 threshold.

Deploy the trained vector model to S3 and build a prediction function that tokenizes plots, maps tokens to vocab IDs, and returns movie genres with scores above 0.5.

Build and deploy a model that reads movie synopses to predict genres, using data extraction, manipulation, a tuned estimator, training, and SageMaker deployment.