Amazon Machine Learning to Predict Responses

Overview

Explore machine learning and computer vision technologies in this conference talk on predicting responses using Amazon Machine Learning. Dive into the concept of foreknowledge in AI and its potential to surpass human capabilities. Learn about tools like Amazon Machine Learning and Amazon Rekognition, and discover how to integrate machine learning algorithms with IoT devices. Gain insights into architecting and developing machine learning algorithms, including the process of obtaining and preparing data, creating models, and enabling real-time predictions. Examine case studies of predictive policing and explore the ethical implications of AI in law enforcement. Understand various machine learning approaches, including binary, multiclass, regression, and unsupervised learning. Discover the potential applications and limitations of machine learning, and get a glimpse into its future developments.

Syllabus

Intro
About Me
General Uses of Machine Learning
When to Use Machine Learning
What Machine Learning is Not
When NOT to Use Machine Learning
Predictive Policing
Study by University of Pennsylvania
United Kingdom (UK) Metropolitan Police
Pittsburgh, Pennsylvania Police Department
Chicago Police Department (CPD)
How Machines Learn
Binary
Multiclass
Regression
Unsupervised Learning
Answerable Questions
Reinforcement Learning
The Machine Learning Process
Obtain Data
Crime Conclusions Based on Analysis
Prepare & Load Data
Prepare Data
Sample Data Observation
Lessons Learned
Load the Data
Create Datasource
Datasource Mapping Exercise
Datasource Created
Sanity Check Data (cont.)
Create (Train) Model (cont.)
Learning Algorithm
Linear Models
Enable the Model for Predictions
Enable Model for Real-Time Predictions
SAM Front End
Main Event Handler
Preparing the Response
Retrieving Crime Prediction
Test Results from Lambda Console
Deleting a Real-Time Endpoint
Things to Note
Machine Blas
Future of Machine Learning