bdsaglam

Midterm Notes & Questions

1. Discriminative vs Generative models: what are the differences? What are the advantages and disadvantages of each?

Discriminative Models ($P(Y|X)$):

• Concept: They learn the boundary between classes. They care about "what differentiates a cat from a dog", not "what makes a dog a dog".
• Goal: Map input $X$ to label $Y$ directly.
• Examples: Logistic Regression, SVM, Neural Nets (standard classifiers).
• Advantages:
• Generally higher accuracy for classification tasks because they focus purely on the decision boundary.

bdsaglam

rules: [
    {
        "description": "hyper + d inserts datetime",
        "manipulators": [
            {
                "from": {
                    "key_code": "d",
                    "modifiers": {
                        "mandatory": [
                            "left_shift",

bdsaglam

DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

bdsaglam

import numpy as np

# Time complexity: O(n), where n is the number of pixels in image
# Space complexity: O(n)
def label_connected_components(image: np.uint8) -> np.int:
    nrow = image.shape[0]
    ncol = image.shape[1]
    component_map = np.full((nrow, ncol), -1, dtype=np.int)
    current_comp_id = 1
    for r in range(nrow):

bdsaglam

//
//  SequenceExtensions.swift
//  Swift6006
//
//  Created by Barış Deniz Sağlam on 10.08.2020.
//  Copyright © 2020 BDS. All rights reserved.
//

import Foundation

bdsaglam

def quotient_remainder(a, b):
    # returns quotient and remainder of a division b
    # a = q*b + r
    q = 0
    while a >= b:
        q += 1
        a -= b
    
    return q, a

bdsaglam

def expm(x, order=10):
    '''
    nth-order Taylor approximation of matrix exponential
    '''
    if order < 1: 
        raise ValueError("order cannot be smaller than 1")
    
    I = torch.eye(x.shape[-1], dtype=x.dtype, device=x.device)
    result = I
    nom = I

bdsaglam

# Credits: https://gist.github.com/nguyenkims
import logging
import sys
from logging.handlers import TimedRotatingFileHandler

FORMATTER = logging.Formatter("[%(asctime)s][%(name)s][%(levelname)s] - %(message)s")
LOG_FILE = "app.log"


def make_console_handler():

bdsaglam

import pathlib
import hydra

hydra._internal.hydra.GlobalHydra().clear()

config_dir = pathlib.Path('/path/to/configs/')
hydra.experimental.initialize(config_dir=config_dir)

cfg = hydra.experimental.compose(config_file='config.yaml', overrides=[])
print(cfg.pretty())

bdsaglam

data/

# Jupyter notebook checkpoints
notebooks/
*.ipynb_checkpoints/

# PyTorch Lightning
*lightning_logs/

# PyCharm files