Modern Artificial Intelligence with Zero Coding

Case study trains chest disease detection with Google Teachable Machine on chest x ray images, classifying healthy, Covid-19, bacteria, pneumonia, and viral pneumonia; builds, trains, and evaluates AI models.

Explore the rise of artificial intelligence in health care, from basics and applications to deep learning in imaging, including lung disease detection and skin cancer screening with Google teachable machines.

Explore how artificial neural networks train and test by randomly initializing weights, splitting data into training and testing sets (80/20), and iterating over epochs to optimize performance and generalize.

Explore key AI lingo such as epochs, batch size, learning rate, and confusion matrix, and learn how training vs testing data, data splits, and weight optimization shape model performance.

Explore how a 4x4 confusion matrix visualizes classifier performance, detailing true positives, true negatives, false positives, false negatives, accuracy, misclassification rate, and type one and type two errors.

Explore visually building and training neural networks with TensorFlow Playground, adjusting layers, neurons, activation, and learning rate for classification tasks, and experimenting with regularization, data splits, and feature engineering.

Explore convolutional neural networks for image data, detailing how convolutions extract features, pooling downsamples, and flattening feeds a dense classifier to recognize digits from 0–9.

Build a simple neural network with Google Teachable Machine to classify chest x-rays into normal, covid 19, viral, and bacterial, then evaluate, export, deploy, test, and compare parameter changes.

Build a four-class image classifier in Teachable Machine for normal, viral pneumonia, bacterial pneumonia, and Covid-19 with 50 images per class and 15 epochs. Export and deploy; test samples.

Explore emotion ai by classifying facial emotions, sadness, happiness, and surprise, using live webcam data, Google Teachable Machine, transfer learning, and evaluating precision, recall, and deployment of emotion models.

Learn to use Google Teachable Machines for live data collection with your webcam to train a model that classifies happy, sad, and surprised facial expressions.

Train the model using Teachable Machine with live webcam data for emotions—happy, sad, surprised—adjusting batch size, learning rate, and epochs, then assess performance in the next lecture.

Test and deploy your Teachable Machine model by using live webcam data, evaluating with a confusion matrix and accuracy, and exporting via TensorFlow.js for a shareable deployment.

Explore classifier model evaluation through confusion matrices, mastering precision and recall, and interpreting true/false positives and negatives, accuracy, and error rate in unbalanced data (cancer example).

Explore off the shelf neural networks, the image net dataset with 14+ million images, and residual networks with skip connections that enable deep architectures up to 152 layers.

Use DataRobot's zero-coding AI to detect cardiovascular disease from patient features - age, height, weight, gender, lifestyle factors, blood pressure, cholesterol, and glucose levels - producing a binary classifier.

Select the target variable and perform exploratory data analysis with Data Robot, viewing distributions and outliers, and apply a three-way train validation holdout split for binary cardio classification.

Train a model in DataRobot by selecting cardio as the target, review cross-validated partitions, and compare models from random forest to XGBoost, noting feature importance like systolic blood pressure.

Explore key binary classification metrics, including the confusion matrix, precision, recall, and accuracy, and learn to interpret the ROC curve and AUC for model evaluation.

Explore model evaluation in DataRobot by comparing precision, recall, and AUC through confusion matrices and roc curves, review multiple trained models, and interpret feature importance and prediction explanations.

Deploy the best extreme gradient boosted trees model with DataRobot in one click, then perform inference and view model explanations, predictions, and deployment health for cardio data.

Discover how XGBoost, extreme gradient boosting, uses an ensemble of weak models and gradient boosted trees for regression and classification, with robust handling of missing data and SageMaker hyperparameter tuning.

Learn how decision trees split data with features and how ensemble boosting methods like Xgboost combine multiple CART models to reduce variance, curb overfitting, and improve robustness.

Explore the extreme gradient boosting framework XGBoost and its scalable, end-to-end tree boosting. Learn sparsity-aware algorithms, weighted quantile sketch, and AWS SageMaker integration for classification and regression.

Explore AWS SageMaker Autopilot's zero-code AI, automating feature engineering, data cleaning, model building, and deployment to predict insurance premiums, applying clustering, recommendations, classification, regression, and time series forecasting in business.

Explore Amazon Web Services as the leading cloud platform, with storage, compute, and SageMaker Autopilot for AI. Understand how data, models, and compute power drive cloud training and deployment.

Discover how Amazon S3 stores data in durable, scalable buckets with region controls, and Amazon EC2 provides on demand compute for training AI models using SageMaker.

Explore AWS SageMaker, a fully managed machine learning workflow that handles labeling, training, tuning, and deployment, while SageMaker Studio with Autopilot enables near-zero coding and automated hyperparameter optimization.

Explore regression metrics for assessing model performance, including mean absolute error, mean squared error, root mean squared error, and the coefficient of determination (R-squared), using residuals, y, and y hat.

Explore how to use AWS SageMaker Autopilot to run experiments on a provided insurance data set, upload data to S3 with a zero coding script, and deploy the best model.

Create an AWS SageMaker AutoPilot experiment in S3 with charges as the target column, letting AutoPilot handle pre-processing, feature engineering, and model tuning across multiple zero-coding trials.

Automate ai ml workflows with aws SageMaker autopilot in studio, train on an s3 csv, pick the best model by mean squared error, and deploy at scale.

Build and deploy a food recognition model with DataRobot, visualize CNN activation maps, and use grad-cam to explain why the AI classifies foods.

Upload your image dataset to DataRobot, train and deploy a four-class AI classifier (fried food, seafood, vegetables and fruits, and dessert), and explore model results with the AutoML leaderboard.

Explore explainable ai with grad-cam visualizations in datarobot, comparing mobilenet v3 and a keras slim residual cnn. Visualize activation maps and heatmaps to localize regions driving predictions.

Explore logistic regression as a probabilistic classifier that uses a sigmoid of a linear model to predict binary outcomes, transforming probabilities into class labels with a threshold.

Explore how bias and variance shape model performance, from simple linear regression to complex polynomial fits, using training and testing data. Find the optimal bias-variance balance for generalization.