IT 3040

1. Machine Learning

   1. AI - Human Intelligence exhibited by machine

   2. Machine Learning - A method to achieve AI through algorithms that can learn from experience to find patterns in a set of data

   3. Neural Networks - A subset of ML techniques inpired by bio systyms, Collectuons of connected units, organized iun layers

   4. Deep learning - A subset of neural networks with more layers and connectuons. Examples include..

   5. Generative AI - A collection of deep learning algorithms that can generate human like content

2. Overview

   1. Machine Learning

      1. A field of study that gives computers the ability to learn without beign explicity programmed.

      2. Arthur Samuel (1959)

   2. Question

      1. IF the checkers program had been allowed to play only 10 games instead of 10s of thousands against it self, how would the performance be affectied?

         1. It wouldve made it worse

   3. Machine Learning Algorithms

      1. Supervised Learning

         1. Rapid advancements. Used in most real world applications

         2. Have the data and the answer

      2. Unsupervised Learning

         1. Have the data but no answer

      3. Recommender Systems

         1. How reccomendations are made to you online based on history, what you bought, ect. Used in streaming, online shopping

      4. Reinforcement learning

         1. Will create a policy that allows the system to operate in some fashion. Think self driving cars. requires a testing enviorment

      5. Practical Advice for applying learning systems

3. Supervised learning

   1. Learns from being given the right answer

   2. Applications include spam detection, speech to text. translations, online advertising

   3. Regression - Predicts an infinite possibilitys

4. Unsuoervised Learning

   1. Data only comes with input x, but not input y. Algorithim has to find structure in the data

   2. Clustering: Group similar data points together

   3. Dimensionality reductuon. Compress data

Supervised Learning

1. Elements of a machine learning Model

   1. The model: Makes predictions

   2. A cost function: The cost function provides a way to quantify how well the ML model is performing. (how accurate it is)

   3. An optimizer algorithm: The optimizer is used to update the algorithm to reduce the error and improve performance

2. Terminology

   1. Training Set: Data used to train the model

   2. x = input var or feature

   3. y = output var or target

   4. m = total number of training examples

   5. (x, y) = a single training example

   6. (x^i, y^i) = the ith training example

   7. x^(2) ≠ x²

3. Process

   1. Training set → learning algorithim → function

   2. features go into the model, and prediction is made (estimated y)

   3. How do we represent f (function)?

      1. With linear regression, F is going to be an equation

      2. f_wb(x) = wx +b

4. Cost Function

   1. w, b: parameters, coefficients, weights

   2. what do w, b do?

8/28/2025

1. Elements of Gradient Descent

   1. Have a cost func

   2. Want Min w,b j(w,b)

2. Outline

   1. Start with some w, b

   2. Kepp changing to reduce j(w, b)

   3. Until we settle at or near a minimum

9/3/2025