Core Technologies Flashcards
AI
A system that can perform one or more human abilities.
Types of AI
- Weak AI
- General AI
- Strong AI
Weak AI
A system used to perform a specific task with high degree of proficiency and lacks general reasoning capabilities.
Examples of weak AI
- Virtual Assistants
- Recommendation systems
General AI
Has the ability to understand, learn and apply knowledge on a wide range of tasks at a human-like level.
Strong AI
Surpasses human capabilities in virtually all fields.
AI Functioning types
- Self learning AI
- Rule based AI
Machine Learning
Capability of self-learning.
How does Machine Learning work?
The creator needs to feed it data for it to study it.
Deep Learning
Is a part of machine learning where neural networks mimic human brains.
Neural Networks
Replica of the human brain using a mathematical formulae.
Neural Pathways in Humans
Connection between Neurons
Neural Pathways in Neural Networks
Connection between mathematical functions that mimic neurons.
Similarities of ML & DL
- Are sub categories and algorithms of AI
Differences of ML & DL
- ML uses creator given data and DL uses neural networks
Types of ML
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
Supervised Learning
Is provided with labelled input variables and corresponding labelled outputs.
Types of supervised learning
- Regression Algorithm
- Classification Leaning
Regression Algorithm
- Predicting numeric outputs based on patterns observed from the features of the labelled inputs and outputs.
Classification Learning
Provision of pre-designed/pre-defined categories so that it will classify data according to the human given criteria.
No final values are created, the inputs are just categorized.
Unsupervised Learning
Provision of unlabeled data and the AI studies the information.
Type of unsupervised learning
- Clustering
Clustering
No provision of pre-defined categories or parameters, the AI needs to group the data based on the similarities of the data.
It is up to the humans to label the cluster created by the AI.
Difference between Classification and Clustering
In classification, the inputs are categorized based on pre-defined categorizes.
In clustering, the inputs are categorized based on self-learnt similarities into unlabeled groups.
Reinforcement Learning
Is when the AI is exposed to the natural environment with no guidance and is asked to make independent study, later human feedback will be given from which the algorithm will change its strategies.
Process of ML
- Raw Data Set
- Preprocess Data
- Validate Data
- Learning Algorithm
- Candidate Model
- Deploy Model
- Golden Model
- Application
Data Pre-Processing
- Data collection
- Data Cleaning
- Data Integration
- Data transformation
- Data Reduction
Data collection
Getting raw data
Data cleaning
- Handling missing values
- Removes duplicates
- Correct data errors
- Removes noise
- Removes unnecessary information
Data Integration
Combining multiple data sets in a unified format.
Data Transformation
- Generating new features from existing data
- Attribute Selection
- Normalizing
Normalizing
Categorizing data into consistent ranges
Data Reduction
- Using data sampling techniques
- Reduces the number of variables whilst preserving diversity
What does Data Science Comprise of?
- Programming
- Mathematics + Algorithms
- Domain Knowledge
Objective of Data Science
By studying data, understanding the relationship between data in a scientific approach.
Main benefit of Data Science
Allows the creation of strategic planning.
