This is an unofficial reading list from Fall 2026 computer science classes at Princeton
I’ve read sections of many of these, but the items marked with a bold ✔️ are ones I’ve found interesting enough to read at length

←Spring 2026

This is an unofficial reading list from Spring 2026 computer science classes at Princeton
I’ve read sections of many of these, but the items marked with a bold ✔️ are ones I’ve found interesting enough to read at length

Fall 2026→

This is an unofficial list of books that are recommended in Spring 2026 computer science classes at Cornell
I’ve read sections of many of these, but the books marked with a bold ✔️ are ones I’ve found interesting enough to read at length

← Fall 2025

📺 Love, Death & Robots, s.4 by Netflix

📺 Moonrise by Netflix

📺 The Electric State by Netflix

🗣️ AI & Productivity: A General Purpose Technology Approach by Jonathan Haskel

🗣️ Geopolitics of Payments: The Dollar, the RMB & Digital Currencies by Martin Chorzempa

🗣️✨🐯Innovation Conference: Going Boldly: From Idea to Impact✨

📺 Frontline: The Rise and Fall of Terrorgram by PBS

🗣️✨Trenton Computer Festival✨ (TCF)

• After Twenty Years, the Maker Movement Is Still Fun by Dale Dougherty
• A Glimpse into the World of Immersive Technology of Otherworld by Schenck & Skovron
• Using Generative AI to Predict Machine Behavior by Perlowin & Alabsi
• Putting Sense into Artificial Intelligence by Edward E. Herceg
• Temperature Logger Using a Low-cost Raspberry Pi Single-Board Computer by Don Arrowsmith
• Using SysML to Model a Hospital Blood Analyzer by Brian Berenbach

🗣️ Will Quantum Computers Solve the World’s Greatest Problems? by 🐯Andrew Houck

This is an unofficial list of books that are recommended in Fall 2025 computer science classes at Princeton
I’ve read sections of many of these, but the books marked with a bold ✔️ are ones I’ve been interested enough in to read at length

←Spring 2025 | Spring 2026→

This is an unofficial list of books that are recommended in Fall 2025 computer science classes at Cornell
I’ve read sections of many of these, but the books marked with a bold ✔️ are ones I’ve been interested enough in to read at length

Spring 2026→

I use these notes to keep track of my thoughts while reading and to record anything that sticks out to me as interesting. The notes are incomplete as I have not yet finished reading the slides

Introduction

• I’ve heard many definitions of machine learning and artificial intelligence, but the ones use by this course are machine learning: “programs that improve with experience, a subfield of artificial intelligence” and artificial intelligence: “programs that demonstrate ‘intelligence’ in some sense”
• The slides contain a fun flowchart: (training data + training output → program / model) + test data → test output

ML basics

• Definitions adapted from notes:
  • Supervised learning: “make predictions from data”
  • Classifier: “a program to predict the correct label of each annotated data instance”
  • Binary classification: a label is one of two possibilities
  • Multi-class classification: a label is one of K possibilities
  • Regression: the label space is the real numbers
  • Feature vector: the input vector of a sample
  • Feature: a dimension of the feature vector
  • Dense feature vector: the number of nonzero coordinates in the feature vector is large relative to the number of features
  • Sparse feature vector: the feature vector consists of mostly zeros
  • Hypothesis class: the set of possible hypothesis functions, the set of hypothesis functions we can possibly learn, “encodes your assumptions about the data set / distribution”
  • No Free Lunch Theorem: “every successful ML algorithm must make assumptions. This also means that there is no single ML algorithm that works for every setting”, “you must make assumptions in order to learn. no Algorithm Works in all settings”
  • Loss function / risk function: “evaluates a hypothesis on our training data and tells us how bad it is”, “tells us how good a model did on an instance”
  • Loss: “the higher the loss, the worse a hypothesis is - a loss of zero means it makes perfect predictions”
  • Zero-one loss: “counts how many mistakes a hypothesis function makes on the training set”
  • Normalized zero-one loss / training error: “the fraction of misclassified training samples”
  • Overfitting: get low error on the training data, but does horribly with samples not in the training data
  • Weak law of large numbers: “the empirical average of data drawn from a distribution converges to its mean”
  • Feature extraction: Selecting “part of instances we deem relevant for predicting output”
  • Model / program / hypothesis: function from input to “label / output we would like to predict”

K-nearest neighbors and the curse of dimensionality

• Definitions:
  • k-NN algorithm: “or a test input, assign the most common label amongst its k most similar training inputs”

  • Bayes optimal classifier: “_you knew P(y

    x), then you would simply predict the most likely label_”

• According to Wikipedia: “[Minkowski distance](https://en.wikipedia.org/wiki/Minkowski_distance) is typically used with p being 1 or 2, which correspond to the Manhattan distance and the Euclidean distance, respectively. In the limiting case of p reaching infinity, we obtain the Chebyshev distance”

I use these notes to keep track of my thoughts while reading and to record anything that sticks out to me as interesting

Course intro, Linux and bash

I use these notes to keep track of thoughts I have during reading, and anything that sticks out to me as interesting. My notes are incomplete as I have not yet finished reading the slides

0. Introduction

• The everything is correctly in quotes in the statement “ ‘Everything’ can be encoded as a sequence of bits ” because continuous things, like real numbers, cannot always be encoded in a finite sequence of bits

• The class continues to use Java as it did when I was a teaching assistant
  The upside is that Java’s verbosity and strong typing make it very explicit what you are doing, and provide a strong foundation of understanding. The downside is that those same features make Java a more difficult first language\ Some universites, like Cornell, use a looser language like Python for the first programming course to get students coding quickly, then Java for the second to learn the deeper details of programming
  I don’t know which system works best—it probably depends on the student

0. Hello, world

• Machine language, natural language, and high-level programming language are contrasted, and it is pointed out that natural language is getting more understandable by computers since I was a teaching assistant for the course

0. Built-in data types

• “ A data type (type) is a set of values and a set of operations on those values ” is a key concept in programming languages

• Java overloading the division symbol (/) to perform both integer and floating point division causes confusion

• The standard warning not to use equality testing (==) with floating point numbers is presented

• Java is statically typed: the types of expressions are determined at compile time, and the compiler will use the type system to catch common errors

• Java will concatenate (+) strings with other types automatically (e.g. using the toString method for objects)

• Explicitly casting floating point numbers to integers rounds down

1. Conditionals

• Local variables declared in a Java code block are not accessible outside the block

• Even if the body of an if statement only has a single statement, it is recommended to put that single statement in a code block, which is good advice.

1. Loops

• Java’s for-each / enhanced-for statements are not explored, because which limits the power of for statements, but avoids talking about Collections this early in the class.

2. Arrays

• The example double[] a = new int[10]; is used to demonstrate a compile-time type mismatch error.
  What’s interesting is that arrays of reference types in Java are covariant, allowing Object[] a = new String[10];. However, this will cause a runtime ArrayStoreException if you try to store anything other than a String in array a.
  When generics (List<T>) were added to Java, they were correctly invariant–List<Object> l = new ArrayList<String>(); will fail to compile with a type-mismatch error.
• I was uncertain if the shuffle algorithm presented was correct, because it was different from the traditional presentation, but apparently it is indeed correct so that was cool to learn

3. Functions

• Functions can only return a single item. Unlike in languages that have tuples built-in, if you want to return more information from a function, you can create a class to hold the information, or you can use the syntactic sugar of record classes
• Java chooses which overloaded function to call based on the arguments and their types, not based on the return type
• The scope of a Java variable is until the end of the block in which it is declared

3. Libraries & Clients

• A sawtooth function is implemented: 2 * (freq*t - Math.floor(freq*t + 0.5))

4. Recursion

• Fun use of M.C. Escher’s Circle Limit IV (Heaven and Hell)
• Dynamic programming can be top-down memoization, or can build the solutions from the bottom-up. Top-down can be simpler to program (just add memoization to existing recursive code), but is usually slower because still involves recursive calls.

4. Performance

• Algorithms topics focus on system-independent effects, while systems topics focus on system-dependent effects like hardware, systems software, and operating system. I always found the latter super-interesting
• Reminder that (n choose k) = n!/k!(n-k)! = (n*…*(n-k+1))/k! This definition often comes in useful
• There used to be confusion over the size of a kilobyte: 10^3 (1,000) or 2^10 (1,024). Now kilobytes are 10^3 (1,000) bytes, while kibibytes are 2^10 (1,024) bytes. The same holds for other prefixes: mega / mebi , giga / gibi …
• In Java, bytes are 8 bits, short : 16 bits, int : 32 bits, long : 64 bits, float : 32 bits, double : 64 bits, char : 16 bit unicode, boolean s use an unspecified number of bits