2.155 - Artificial Intelligence and Machine Learning For Engineering Design

2.155/2.
156 Artificial Intelligence and Machine Learning for

Engineering Design
Instructor Faez Ahmed

[email protected]
Office hours: Monday noon - 1 pm in 3-435A
Class MW 1:00 PM – 2:30 PM in 4-145
Subject Undergraduate (2.155)

Graduate (2.156)
12 units
3-0-9
Course [TA] Lyle Regenwetter

Assistant [email protected]
TA office hours: TBA
Description: In this course, students will delve into the applications of Artificial
Intelligence and Machine Learning in engineering design, focusing on
the creation of new products and addressing engineering design
challenges. Emphasis will be on the practical implementation of
advanced machine learning and optimization strategies. Participants will
be involved in a term project, employing these AI techniques to address
complex problems of their choice. Additionally, there will be hands-on
exercises in machine learning tailored to specific engineering issues,
such as robotics, aircraft, structures, and metamaterials. The curriculum
also incorporates discussions on research papers, facilitated by
students. This course is designed for engineering students with
preliminary knowledge in programming and machine learning, aiming to
apply these tools across diverse engineering landscapes.
Prerequisites 2.086, 6.100A, or permission of instructor.

The class does not expect you to have a background in machine
learning. However, familiarity with programming skills is recommended.
List of topics ● Introduction to Python

● Data Visualization
● Single and Multi-objective Design Optimization
● Similarity, Norms, and Feature Engineering
● Supervised Learning: Classification Methods
● Supervised Learning: Regression Methods
● Cross-Validation
● Unsupervised Learning: Clustering Methods
● Dimensionality Reduction
● Neural Networks: MLPs and Convolutional Neural Nets
● Self-Supervised Learning
● Deep Generative Models: Generative Adversarial Networks
● Deep Generative Models: Diffusion Models
● Applications of Generative Models for Design Synthesis
● Computational Creativity and Novelty Detection
● AutoML
● Natural Language Processing Introduction
● Transformers: BERT, GPT
● Interpretability of Neural Networks
● Statistical Graph Learning for Complex Systems
● Graph Neural Networks
● Ethics in Machine Learning
● Challenges in Data-Driven Design Applications
Term project Engage in a comprehensive term project, utilizing the skills and
techniques learned throughout the course. Students are encouraged to
tailor their projects around personal or research interests. Conclude the
term by showcasing your work through an end-of-semester poster
presentation.
Challenge Join teams to tackle AI-centric design challenges. Engage with varied
problems facets of machine learning and design through each problem, ranging
from optimization to deep generative models. Submissions should
include your solutions, relevant code, and a reflective report on your
approach.
Research paper Some classes will start with a 30-minute chat led by a student about a
discussion design research paper. For these classes, you need to write a short
review of a paper you pick. This review should have a summary and
your thoughts about the paper, plus any questions you think would be
good to talk about. Make sure to give this in before the class starts.
Grading Grading will be based on projects, challenge problems, and paper

presentations. The grading breakdown and milestones will be discussed
in the first lecture. Students taking graduate version (2.156) complete
additional assignments.
Programming We will use the Python programming language for all assignments in
this course. For students who have never used Python, introductory
tutorials and starter code will be provided for most of the problems and
demos. While other programming languages such as MATLAB, Julia,
and R are permissible for assignments, primary support and resources
will be provided for Python.
Recommended There is no single textbook that covers all of the topics we will discuss in
Text Books this course. The recommended books listed below may be useful for a
deeper understanding of the topics.
● Ulrich, Karl T. & Eppinger, S. D. & Yang, M. Product design and

development. Tata McGraw-Hill Education.
● Kevin Murphy, Machine Learning: A Probabilistic Perspective,
MIT Press
● Aaron Courville, Ian Goodfellow, and Yoshua Bengio, Deep
Learning, MIT Press
● Tom Mitchell, Machine Learning, McGraw Hill
Office hours To improve students’ familiarity with Python and its libraries, we will
conduct weekly TA office hours to cover student needs for Python
knowledge and machine learning libraries. These office hours will be
used to explain the challenge problems when they first come out,
answer any questions related to the problems, and provide an
opportunity for open Q&A to any questions that were not addressed
during lectures. Attendance is optional.
Previous Year The class was offered in Fall 2021 and Fall 2022 as 2.s997. The
Projects previous projects covered a diverse range of applications, from medical
design to energy systems, with pioneering methods like generative
adversarial networks and symbolic optimization. This showcases the
course's extensive reach into contemporary engineering topics and
innovative computational approaches.
List of projects from 2022:
● Learning Deformable Point Cloud Correspondences in Medical

Ultrasound Data
● Motorsports Cooling Optimization
● Optimized Torsion Spring for Rehabilitative Exoskeleton

● Nozzle Constraints Aware Data-Driven Topology Optimization
● Predicting Loads on a Decentralized Grid
● Wind Farm Model and Wake Steering Optimization with PyWake
● Using Machine Learning to Optimize Black Box Simulations
● Latent Space Design Exploration of Complex Structures
● Temperature and Density Measurements Using Optical Spectroscopy

and Machine Learning in Inhomogeneous Non-Optically Thin Plasmas
● Implementing Natural Language Processing to Assess the Design and

Performance of the iPhone SE
● Machine Learning for Inverse Kinematics
● Classifying the Impact Performance of Truss-Based Lattices
● Efficient Meshing Scheme for the AMORE Mesh
● Data-Driven Modeling of McKibben Actuator Dynamics
● Bike Design Analysis and Classification Using AutoML
List of projects from 2021:
● Data-Driven Design of Customized Prosthetic Feet
● Multi-Objective Optimization of Ventricular Catheters Using Genetic

Algorithm
● Symbolic Regression and Variational Autoencoder Generation of

Metamaterial Unit Cells
● When Machine Learning Meets IC Engine: ML Application on Gas

Flow Pattern in Piston Ring Pack System
● Evaluation Method on 3E+S for Energy Mix Options in Power

Generation Market Using GA Algorithm
● Preliminary Benchmark Study of Algorithm Configuration, Selection,

and Evaluation for Data-Driven Modeling of Experimentally-Derived
Datasets
● Robot Exploration Initialization Optimization

● Using Segmented Images to Improve Bicycle Image Reconstruction
with Variational Autoencoder
● Using Machine Learning Model to Optimize Peak Load and Heat

Stress Under Heat Waves for Buildings in Various Dimensions by
Adjusting Fan Speed and Thermostat Setpoint
● Multi-Objective Optimization for Strength-Ductility Tradeoff Using

Deep Learning
● Topology Optimization Using CNN and Conditional GAN
Below are a few photos from the final poster presentations in 2022 (top) and 2021 (bottom).

2.155 - Artificial Intelligence and Machine Learning For Engineering Design

Uploaded by

Copyright:

Available Formats

2.155 - Artificial Intelligence and Machine Learning For Engineering Design

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2.155 - Artificial Intelligence and Machine Learning For Engineering Design

Uploaded by

Copyright:

Available Formats

2.155/2.

156 Artificial Intelligence and Machine Learning for

Instructor Faez Ahmed

Class MW 1:00 PM – 2:30 PM in 4-145

Subject Undergraduate (2.155)

Course [TA] Lyle Regenwetter

Prerequisites 2.086, 6.100A, or permission of instructor.

List of topics ● Introduction to Python

Grading Grading will be based on projects, challenge problems, and paper

● Ulrich, Karl T. & Eppinger, S. D. & Yang, M. Product design and

List of projects from 2022:

● Learning Deformable Point Cloud Correspondences in Medical

● Motorsports Cooling Optimization

● Optimized Torsion Spring for Rehabilitative Exoskeleton

● Predicting Loads on a Decentralized Grid

● Wind Farm Model and Wake Steering Optimization with PyWake

● Using Machine Learning to Optimize Black Box Simulations

● Latent Space Design Exploration of Complex Structures

● Temperature and Density Measurements Using Optical Spectroscopy

● Implementing Natural Language Processing to Assess the Design and

● Machine Learning for Inverse Kinematics

● Classifying the Impact Performance of Truss-Based Lattices

● Efficient Meshing Scheme for the AMORE Mesh

● Data-Driven Modeling of McKibben Actuator Dynamics

● Bike Design Analysis and Classification Using AutoML

List of projects from 2021:

● Data-Driven Design of Customized Prosthetic Feet

● Multi-Objective Optimization of Ventricular Catheters Using Genetic

● Symbolic Regression and Variational Autoencoder Generation of

● When Machine Learning Meets IC Engine: ML Application on Gas

● Evaluation Method on 3E+S for Energy Mix Options in Power

● Preliminary Benchmark Study of Algorithm Configuration, Selection,

● Robot Exploration Initialization Optimization

● Using Machine Learning Model to Optimize Peak Load and Heat

● Multi-Objective Optimization for Strength-Ductility Tradeoff Using

● Topology Optimization Using CNN and Conditional GAN

You might also like