L:11 AI & ML

Key Points about Supervised Learning

- Each data tagged with the correct label. - Predict future outcomes based on past data - Requires an input and output variables - Training and testing the model.

Weak AI: Artificial Narrow Intelligence (ANI)

- requires training to perform a specific task. - drives most of the AI that surrounds us today. - E.g. Apple's Siri, Uber's self-driving, Facebook's facial recognition ,and Google Translate

For training the model, data set is split into 2 segments:

- training data - testing data.

Test set

A set of unseen data used only to assess the performance of a fully-specified classifier.

Evaluate "Tune" the model

A confusion matrix is used to evaluate the trained model, i.e., tell how well the trained model is.

Choose the model

Select the model that will be best for the type of your data. The goal is to train the best performing model possible using the pre-processed data.

Artificial Intelligence (AI)

Systems or machines that mimic human intelligence to perform tasks and can iteratively improve themselves based on the information they gather

Features

The input variables that we give to our machine learning models

True positives (TP)

These are cases in which we predicted TRUE, and our predicted output is correct.

True negatives (TN)

We predicted FALSE, and our predicted output is correct.

False negatives (FN)

We predicted FALSE, but the actual predicted output is TRUE.

False positives (FP)

We predicted TRUE, but the actual predicted output is FALSE.

Training set

a material through which the computer learns how to process information and tune the parameter of classifier.

Clustering

a problem in which a set of inputs is to be divided into groups. Example: identifying sub-groups of Patients with chronic cough (No ICD codes for chronic cough)

Classification

a problem in which the target variable is categorical ▪ Examples: classify an email as "spam" or "not spam", classify an x-ray of patient as having "pneumonia" or "no pneumonia". ▪ Example of classification algorithms: K-Nearest Neighbor, Support Vector Machine

Regression

a problem in which the target variable is continuous ▪ Example: Predicting prices of a house given the features of house like size, price. ▪ Example of regression algorithms: Linear Regression, Decision Tress/Random Forest

Machine Learning (ML)

a sub-field of AI that focuses on building system, which has the ability to learn without being explicitly programmed. It focuses on using data and algorithms to simulate the way humans learn and gradually improve its accuracy based on data they consume.

Deep Learning

a subset of machine learning that learn the representation of data using a hierarchy of multiple layers that mimic the neural network of human brain.

AI In Saudi Arabia

was established in 2019

The Challenges of Adopting Artificial Intelligence

▪ Building Trust ▪ Lack of skilled professional ▪ Data Security ▪ Software Malfunction ▪ Data Scarcity ▪ Algorithm Bias

Deep Learning Application

▪ Cancer tumor detection ▪ Image coloring ▪ Object detection

he supervised learning is categorized into 2 other categories

▪ Classification ▪ Regression

pre-processing techniques include:

▪ Conversion of data ▪ Ignoring the missing values ▪ Filling the missing value ▪ Outliers' detection

Why AI Now?

▪ More Computing Power ▪ Availability of More data ▪ Better Algorithms

AI In Saudi Arabia develop several AI-based program:

▪ Tawakkalna ▪ Tabaud ▪ Boroog ▪ Ehsan

Types of artificial intelligence

▪ Weak AI ▪ Strong AI

Outliers' detection

some error data is presented in data set that deviates drastically from other observations. For example: mistyping

Feature Extraction

The process of choosing relevant features for your machine learning model based on the type of problem you are trying to solve Example: the model decide which student needs extra tutorial sessions

Gathering data

The process of gathering data depends on the type of project. The data set could be collected from database, files, or sensors. The collected data can't be used directly to perform analysis

AI drive the Kingdom's advancement in artificial intelligence innovations:

• orchestrating AI research • developing AI solutions • enhancing AI education

Key Points about Unsupervised Learning

- AI system is presented with unlabeled, un-categorized data and the system's algorithms act on the data without prior training to discover patterns. - he system identifies hidden features from the input data provided. - Example of unsupervised learning is "Clustering".

A confusion matrix has 4 parameters

- True positives - True Negatives - False Positives - False Negative

Strong AI: Artificial General Intelligence (AGI)

- a theoretical form of AI where a machine would have an intelligence equaled to humans. - has a self-aware consciousness that has the ability to solve problems, learn, and plan for the future. - urpass the intelligence and ability of the human brain. - Not exist yet

Machine Learning Processes

1- Gathering data 2- Data Pre-processing 3- Choose the model 4- Evaluate the model 5- Deploy the mode

ML Types

1- Supervised Learning 2- Unsupervised Learning

Accuracy

= (TP +TN) / (TP+TN+FP+FN)

Deploy the mode

After the evaluation model phase is done, the resulted " trained" model can be utilized.

Unsupervised Learning

Discovering patterns in unlabeled data Example: cluster similar documents based on the text content

Supervised Learning

Learning with a labeled training set Example: email spam detector with training set of already labeled emails

Conversion of data

ML models handles only numeric features. Converting categorical and ordinal data into numeric

Data Pre-processing

Most important step that help building machine learning models more accurately. A process of cleaning the raw data and converts it into a clean data set.

Ignoring the missing values

Removing the row or column of data that has missing value.

Artificial Intelligence (AI)

a set of computer science techniques that allows computer software to learn from experience, adapt to new inputs and complete tasks that resemble human intelligence

Filling the missing value

fill the missing data manually, e.g., with mean, median or highest value.