Loading...

Are you interested in truly understanding what is happening inside the black box of contemporary machine learning methods? Do you want to have a clear sense of what the tradeoffs are between different methods, what the pitfalls are, and how to avoid them? Do you want to contribute to decision-making about machine learning applications?  Due to advances in the statistical and data science domains; in the capacity of affordable computer hardware; and in the availability of useful data, machine learning has broken out of the purely theoretical world and is now permeating businesses at a fast pace and on a global scale. This has triggered a resulting surge in the incorporation of machine learning methodology in successful academic pursuits.  Today, professionals, academics, and anyone else with the goal of using data to inform decision-making serves to benefit from a clear understanding of the collection of powerful machine learning methods driving these changes. This course focuses on developing an intuitive and memorable understanding of how these methods work. While the logic behind each method will be covered thoroughly, it will be derived and justified in plain English and with clear examples such that anyone with a background in algebra can follow. The course will also cover how these various methods are applied to solve problems and inform decisions, emphasizing what can go wrong and how to avoid such mistakes.

Course Details

Module 1: Intro to Machine Learning.
This module will cover what machine learning is at a high level and how the specific practice of machine learning tends to fit in to larger standard processes and organizational charts within organizations. This module will also cover the important differences between structured and unstructured data, classification and regression, as well as supervised and unsupervised learning. Finally, this module will include brief introductions to the specific methods covered in the rest of the course, explaining why they were selected and how they relate to each other.

Module 2: Gradient Descent.
This module will cover how to fit a line to data in the simple case of a single input variable function, and then it will expand to explore the case of multi-dimensional input data. Both batch and stochastic gradient descent will be covered. This module will also explore some general techniques for assigning accuracy to models and revising them accordingly. Participants will engage in exercises covering each of these topics and will hear examples of how gradient descent has been used in various case studies.

Module 3: Random Forest.
This module will build up the intuition for the random forest method through three separate lectures, each with its own exercise. Participants will learn how decision trees work, how model accuracy can be simply improved by adjusting averages, and how sample grouping can be programmatically optimized. The module will then cover how to turn the resulting tree into a random forest, how to put the forest to use both for regression and classification work, and how to measure and refine accuracy. This module will also consist of several individual and group exercises. Example implementations of random forest will then be shared.

Module 4: Introduction to Support Vector Machines.
Introduction to Support Vector Machines. This module will start with the simple case of finding the best binary classification model for two-dimensional data in both the supervised and unsupervised setting. From there, the module will cover conceptually how kernel functions are used to map input variables into higher dimensional space to enable linear separating. This module will cover several case studies of how support vector machines have been useful in industrial applications.

Module 5: K-Means and K-Nearest Neighbors.
This module will build on the first exercise from the previous module, introducing unsupervised K-Means clustering and then moving into K-Nearest Neighbors for supervised classification and regression. Special attention will be devoted to distinguishing between these two oft-confused methods. Participants will engage in an evolving group exercise to prove the concepts to themselves. Several examples from the industry will be shared. Module 6: Introduction to Artificial Neural Networks. This final module will cover the architecture and steps of artificial neural networks, including definitions and examples of common terms (“deep learning”, convolutional, etc). The module will derive the overall cost function as well as forward and back propagation, including the various options for activation functions, but participant application will be limited to rudimentary and conceptual understanding. The exercises in this module will have participants build nodes and layers, apply an activation function, make a prediction, and then explore how back propagation would then update the model to repeat the process.

Module 6: High Level Introduction to Artificial Neural Networks and Deep Learning

This optional bonus module will cover the high level architecture and basics of artificial neural networks (ANN) and deep learning. Participants will learn how ANNs resemble a working model of biological neural networks, how the nodes that compose an ANN function and communicate, how activation functions tailor the network to particular categories of use cases, as well as how error measuring, forward propagation, and back propagation work. Entire graduate programs are devoted to covering the theory and application of ANN, so we will be unable to cover the topic to an extent lending thorough intuitive understanding the way that other topics are covered in this course.

  • Innovators within the firm, looking for ways to disrupt existing best practices
  • Leaders responsible for managing teams through pursuit of innovation to overcome strategic challenges
  • Professionals with interest in developing new understanding and actionable insights from their customers or to expand their customer base into new demographics
  • Entrepreneurs exploring markets for new venture opportunities

Prerequisites for this course

1. You must be comfortable with basic computer interaction.

  • Web browsing (search engines, tabs, scrolling, refreshing)
  • Copying and pasting content using your operating system's clipboard
  • Opening new windows in programs
  • Checking and sending email
  • Using Microsoft Excel and Word
  • Using Zoom to join video calls

2. No coding background is necessary. While examples of various machine learning methods will be delivered in the form of pre-written code that you can execute, you will not be expected to understand or write such code. The purpose of this course is to build a strong understanding of the theories and processes enabling various machine learning methods as well as how those methods are applied.

3. You must also be comfortable with high school math by the time the course begins. Note, if you need refreshers on any of the below, you are still encouraged to sign up for this course. If you do, you will receive resources that help you solidify your understanding before the course commences. Topics to be familiar with besides basic arithmetic include:

  • Powers
  • Order of operations
  • Functions
  • Graphing (no trigonometry or calculus)
  • Maintain a robust understanding of how popular and powerful machine learning methods work
  • Tell the difference between buzzwords and coherent machine learning plans
  • Evaluate machine learning approaches for comparative advantages
  • Put together successful strategies for utilizing machine learning methods

If you are going on to learn software engineering or computer science, you can tackle the coding side of implementing of data science methods already knowing how the machine learning methods work

What Will Set Me Apart?

  • Talk technical with data and computer scientists while maintaining an ability to implement through others
  • Gauge what could be possible with machine learning, how long it might take, and what resources it might require
  • Propose clear and articulate machine learning protocols and plans in your organization
  • Prevent time and resources from being wasted on ineffective or inefficient machine learning endeavors

Financial assistance information can be found on Rice Financial Aid's Visiting and Continuing Education Students page.

The estimated Cost of Attendance (COA) is $950.00

Due to the high demand for courses, registrations are considered final as of 10 working days before class starts. No refunds will be issued after these dates and credits will not be given for future classes. No refunds will be granted for participants who miss a portion of a program. Refund requests before the deadline are subject to a 10% processing fee and must be made in writing to gscs@rice.edu. If books have been issued, the cost of the books and any shipping fees will be deducted. Refunds for credit card payments will be processed as credits to the accounts from which they were paid and may not appear as a credit until the following statement. Refunds for enrollments paid by check take up to four weeks to be processed and mailed by the Rice University accounting office. There is a $30 charge for any check returned for insufficient funds.

An individual may be able to claim a tax deduction or a tax credit for the cost of attending this continuing education program. However, we do not issue 1098-Ts. Non-academic credit programs such as continuing education courses do not require a 1098-T per IRS guidelines. Please keep a copy of your receipt and consult with your tax advisor about whether your payment meets the regulations that apply to tax deductions and tax credits. The University’s Federal Tax ID Number is 74-1109620

Loading...

Thank you for your interest in this course. Unfortunately, the course you have selected is currently not open for enrollment. Please complete a Course Inquiry so that we may promptly notify you when enrollment opens.