Regularization & Optimization
Regularization prevents overfitting by penalizing model complexity, while advanced optimizers like AdamW and learning rate schedules improve convergence and generalization in neural network training.
Notes
Learning notes, reading summaries, and quick references
Image Classification with Linear Classifiers
Linear classifiers use learned weight matrices and biases to assign class scores, enabling fast inference but only handling linearly separable data.
Embedded Communications
Serial communication suits microcontrollers due to fewer pins. Asynchronous UART uses baud rate and start/stop bits, while synchronous I2C and SPI rely on a clock line. Distinction: UART uses baud rate, I2C address, SPI slave select.
Digital Clock
Interrupts trigger ISRs when IF, IE, and GIE bits are set. Timers count internal clock ticks; counters count external events. A 4-digit display uses rapid multiplexing to create persistence of vision.
Overview & Digital I/O
Embedded systems are dedicated, resource-limited devices. Microcontrollers integrate CPU, memory, and I/O on one chip, offering low cost, small size, and easy updates—ideal for most embedded applications.
Symbol Table
A symbol table stores identifier information in a compiler, supporting declaration tracking, type checking, scope management, and efficient lookup for code generation.
Parsing Table
Explains how to construct LR(0) Action and Goto tables using items, closures, augmented grammar, and state transitions for parsing.
LR(0) Parsing
LR(0) parsing uses a state stack and action/goto tables to shift input symbols or reduce handles, determining whether to accept or report syntax errors.
Bottom-Up Parsing
Bottom-up parsing constructs a syntax tree from input tokens by repeatedly reducing right-hand sides of grammar rules to their left-hand sides using a parse stack.
FIRST & FOLLOW set
FIRST sets identify possible starting terminals of derivations, while FOLLOW sets determine terminals that may appear immediately after a non-terminal in parsing.