Applied Machine Learning
Lecture Notes
January 13, 2026
Contents
1 Lecture 1: Introduction and Embedded Systems 2
1.0.1 Machine Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.0.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Lecture 2: Smart Systems and Applications 12
2.0.1 Decision Making . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.0.2 Reconfigurable Platform . . . . . . . . . . . . . . . . . . . . . . . 15
2.0.3 ESP-S3 EYE: SW ESP-DL . . . . . . . . . . . . . . . . . . . . . . 20
3 Lecture 3: Introduction to Machine Learning 21
4 Assignments 33
4.1 Python Reminder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.1.1 Printing Hello World . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.1.2 Basic Arithmetic Operations . . . . . . . . . . . . . . . . . . . . . 33
4.1.3 List Manipulation Operations . . . . . . . . . . . . . . . . . . . . 33
4.1.4 Understanding TensorFlow Constant Code . . . . . . . . . . . . . 33
4.1.5 Basic Operations Using TensorFlow . . . . . . . . . . . . . . . . . 34
4.2 Assignment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.3 Assignment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.4 Senior Assignment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.5 Senior Assignment 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.6 Senior Assignment 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5 Lab Summaries 43
5.1 Lab 1 – ESP32 Setup, Image Pipeline & Embedded Software Architecture 43
5.2 Lab 2 – Dataset Creation & Training a CNN for Image Classification . . 43
5.3 Lab 3 – Quantization & Embedded Machine Learning Constraints . . . . 44
5.4 Lab 4 – End-to-End Embedded Object Recognition System . . . . . . . . 45
5.5 Overall Course Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . 45
1
,1 Lecture 1: Introduction and Embedded Systems
Learning Objectives
• Understand embedded hardware/software architecture within ML systems
• Design AI/ML applications
• Apply AI/ML concepts on embedded devices
• Implement embedded ML on real devices
• Understand how AI, ML, and big data enhance business processes
• Strategically implement AI and manage AI governance
• Develop embedded AI projects
Overview of AML
• ML concepts for building AI applications
• Software and hardware architecture of embedded systems
• Designing Tiny AI
• Labs: hardware-based ML prototypes (ESP-EYE, Arduino, Raspberry Pi)
• Student AI/ML project presented at end of course
Topics Covered
• General Introduction
• Embedded World: reminders
• AI and AI on the Edge
• AI frameworks and architecture
• Smart systems & AI applications
Embedded Systems: Definitions
General Definition: “Any sort of device which includes a programmable computer but
itself is not intended to be a general-purpose computer.” - Wayne Wolf
Definition 1:
• A combination of hardware and software forming part of a larger machine
• Example: a microprocessor controlling an automobile engine
• Runs on its own without human intervention
• May be required to respond to events in real time
2
, Definition 2:
• Special-purpose computer system
• Designed to perform one or a few dedicated functions (sometimes real-time)
• Contains sensors, actuators (and its control loop)
• Embedded as part of another system
Characteristics of Embedded Systems
1. Dependable
• Reliability: R(t) = probability of the system working correctly, provided that it
was working at t=0
• Maintainability: M(d) = probability of the system working correctly for d time
units after an error occurred.
• Availability: probability of the system working at time t
• Safety: No harm is to be caused
• Security: confidential and authentic communication
2. Efficient
• Energy efficient
• Code-size efficient (especially for systems on a chip)
• Run-time efficient
• Weight efficient
• Cost efficiency
3. Many must meet real-time constraints
• A real-time system must react to stimuli from the controlled object (or the operator)
within the time interval dictated by the environment.
• For real-time systems, right answers arriving too late (or even too early) are wrong
Embedded vs General Purpose Computing
Embedded Systems
• Few applications that are known at design-time.
• Not programmable by the end user. (?)
• Fixed run-time requirements (additional computing power not useful).
• Criteria: cost, power consumption, and predictability
3
, General Purpose Computing
• Broad class of applications.
• Programmable by the end user.
• Faster is better.
• Criteria: cost and average speed
Application Areas
• Automobiles: engine management, ABS, airbags
• Buildings: elevators, lighting, security
• Agriculture: feeding/milking systems
• Space: satellites
• Medical: pacemakers, monitoring
• Office: printers, fax
• Tools: multimeter, GPS
• Banking: ATMs
• Transportation: planes, trains, signaling systems
Evolution of Embedded Systems
From transistor circuits → logic gates (VHDL) → processors → processor-based circuits
→ complex network-based embedded systems with OS and applications.
4
Lecture Notes
January 13, 2026
Contents
1 Lecture 1: Introduction and Embedded Systems 2
1.0.1 Machine Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.0.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Lecture 2: Smart Systems and Applications 12
2.0.1 Decision Making . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.0.2 Reconfigurable Platform . . . . . . . . . . . . . . . . . . . . . . . 15
2.0.3 ESP-S3 EYE: SW ESP-DL . . . . . . . . . . . . . . . . . . . . . . 20
3 Lecture 3: Introduction to Machine Learning 21
4 Assignments 33
4.1 Python Reminder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.1.1 Printing Hello World . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.1.2 Basic Arithmetic Operations . . . . . . . . . . . . . . . . . . . . . 33
4.1.3 List Manipulation Operations . . . . . . . . . . . . . . . . . . . . 33
4.1.4 Understanding TensorFlow Constant Code . . . . . . . . . . . . . 33
4.1.5 Basic Operations Using TensorFlow . . . . . . . . . . . . . . . . . 34
4.2 Assignment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.3 Assignment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.4 Senior Assignment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.5 Senior Assignment 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.6 Senior Assignment 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5 Lab Summaries 43
5.1 Lab 1 – ESP32 Setup, Image Pipeline & Embedded Software Architecture 43
5.2 Lab 2 – Dataset Creation & Training a CNN for Image Classification . . 43
5.3 Lab 3 – Quantization & Embedded Machine Learning Constraints . . . . 44
5.4 Lab 4 – End-to-End Embedded Object Recognition System . . . . . . . . 45
5.5 Overall Course Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . 45
1
,1 Lecture 1: Introduction and Embedded Systems
Learning Objectives
• Understand embedded hardware/software architecture within ML systems
• Design AI/ML applications
• Apply AI/ML concepts on embedded devices
• Implement embedded ML on real devices
• Understand how AI, ML, and big data enhance business processes
• Strategically implement AI and manage AI governance
• Develop embedded AI projects
Overview of AML
• ML concepts for building AI applications
• Software and hardware architecture of embedded systems
• Designing Tiny AI
• Labs: hardware-based ML prototypes (ESP-EYE, Arduino, Raspberry Pi)
• Student AI/ML project presented at end of course
Topics Covered
• General Introduction
• Embedded World: reminders
• AI and AI on the Edge
• AI frameworks and architecture
• Smart systems & AI applications
Embedded Systems: Definitions
General Definition: “Any sort of device which includes a programmable computer but
itself is not intended to be a general-purpose computer.” - Wayne Wolf
Definition 1:
• A combination of hardware and software forming part of a larger machine
• Example: a microprocessor controlling an automobile engine
• Runs on its own without human intervention
• May be required to respond to events in real time
2
, Definition 2:
• Special-purpose computer system
• Designed to perform one or a few dedicated functions (sometimes real-time)
• Contains sensors, actuators (and its control loop)
• Embedded as part of another system
Characteristics of Embedded Systems
1. Dependable
• Reliability: R(t) = probability of the system working correctly, provided that it
was working at t=0
• Maintainability: M(d) = probability of the system working correctly for d time
units after an error occurred.
• Availability: probability of the system working at time t
• Safety: No harm is to be caused
• Security: confidential and authentic communication
2. Efficient
• Energy efficient
• Code-size efficient (especially for systems on a chip)
• Run-time efficient
• Weight efficient
• Cost efficiency
3. Many must meet real-time constraints
• A real-time system must react to stimuli from the controlled object (or the operator)
within the time interval dictated by the environment.
• For real-time systems, right answers arriving too late (or even too early) are wrong
Embedded vs General Purpose Computing
Embedded Systems
• Few applications that are known at design-time.
• Not programmable by the end user. (?)
• Fixed run-time requirements (additional computing power not useful).
• Criteria: cost, power consumption, and predictability
3
, General Purpose Computing
• Broad class of applications.
• Programmable by the end user.
• Faster is better.
• Criteria: cost and average speed
Application Areas
• Automobiles: engine management, ABS, airbags
• Buildings: elevators, lighting, security
• Agriculture: feeding/milking systems
• Space: satellites
• Medical: pacemakers, monitoring
• Office: printers, fax
• Tools: multimeter, GPS
• Banking: ATMs
• Transportation: planes, trains, signaling systems
Evolution of Embedded Systems
From transistor circuits → logic gates (VHDL) → processors → processor-based circuits
→ complex network-based embedded systems with OS and applications.
4
