AI.md

Back to Questions

AI 🤖

Term	Definition
Data	Collection of Facts, Informations or Observations
Data Design	How you organize informations and data.
Data Strategy	Management of People, Prices and Tools used in Data Analysis.
Data Ingestion	Unstructured data extracted from multiple sources and prepared for Analytics, Insights or Model building.
Data Analytics	Collect, Transform and Organize Data in order to make Future Prediction or Forecasting.
Data Visualization	Graphical Representation of Information.
Data Modeling	Creating a Model for Reporting and Forecasting Business Decisions.
Data Analysis	Collect, Transform and Organize Data in order to Draw Conclusion from the past Data.
Data Science	Science of extracting knowledge and meaningful insights from structured and unstructured data.
Business Intelligence	Using data to make smart decisions.
Forecasting	Making Predictions based on Past and Present Data.
Training Data Set	Data used to Train the Model ( Allow Algorithm to Learn from Data )
Validation Data Set	Data used to Select the Best Model ( Optimize Algorithm and Hyperparameter Settings )
Test Data Set	Data used to provide an Evaluation
Predictive Modeling	A Process that uses Data and Statistics to Predict outcomes with Data Models
Machine Learning | ML	Field of Science that gives Computer the Ability to Learn from Data without being Explicitly Programmed
Artificial Intelligence	Simulation of Human Intelligence in Machines to Think like Humans and Mimic their Actions
Deep Learning	ML inspired by Human Brains like Detect Object, Recognize Speech, Translate Speech and Make Decisions
Neural Network	Simulation of Human Brain in Computers to Think like Humans to make Decisions and Mimic the Actions
NLP	Science that gives Machine the Ability to Read, Understand and Communicate through Human Language
Computer Vision | CV	Field of AI that Trains Computers to Understand the Visual World like a Human Eye
Internet of Things	Network of Physical Objects embedded with Sensors, Connecting and Exchanging Data with each others
Data Mining	Turn Raw Data into Useful Information, Discovering Patterns in Large Dataset
Cloud	A Place to Keep Data Online, rather than Computer Hard Drive.
Ecosystem	Various elements interacts in order to Produce, Manage, Store, Organize, Analyze and Share Data.
Data Driven Decision Making	Using Facts to Guide Business Strategy.
Analytical Skills	Qualities and Characteristics associated with Solving Problems using Facts.

---

Data Science
Term	Definition
Machine Learning	Primarily focus on predictions based on known properties of data.
Data Mining	Focus on discovery of past unknown properties of data.
Optimization	Focus on improving the predictions.

Data Mining : Finding patterns in historical data and then use those patterns on current data to make future predictions.

Business Analytics
Term	Definition
Predictive Analytics	Analyze past trends in order to predict the liklihood of future outcomes.
Descriptive Analytics	Track and analyze existing data in order to identify new patterns.
Prescriptive Analytics	Recommend actions based on prior performance.

AI | Elements of AI | Building AI | Real World Applications | Data Science Elite

AI > ML > DL

AI | Artificial Intelligence

• Computers perform the task that normally requires Human Intelligence.
• Mimic the Intelligence or Behavioural patterns of humans.
• Ability to learn and reason like humans.
• Make smart computers like human to solve complex problems.
• Deals with Structured | Semi Structured and Unstructured Data.

Types :

1. Weak AI ( Narrow AI ) | Good in One Task.
2. General AI ( Computer as Smart as Human )
3. Strong AI ( Super AI ) Reduce Human Error | 24 x 7 Availability | Digital Assistance | Fast and Accurate Decisions.

Machine Learning | ML

• Subset of AI
• Art of extracting knowledge from data.
• Computers learn from data and act like humans.
• Computers learn from past data without being explicitly programmed.
• Find patterns and relationship between Features and Labels
• Improve automatically through experience.
• A set of algorithms that allows computers to learn from data without being explicitly programmed.
• Deals with Structured and Semi Structured data.

Types :

1. Supervised Learning
2. Unsupervised Learning
3. Semisupervised Learning ( Small amount of Labeled Data + Large Amount of Unlabeled Data )
4. Reinforcement Learning

Applications :

1. Predictions
2. Classifications
3. Segmentation
4. Recommendation System
5. Google Search Algorithm
6. Facebook Auto Friend Tagging Suggestions

DL | Deep Learning

• AI Function that mimics the workings of Human Brain in Processing Data
• Use in Detecting Objects, Recognizing Speech, Translating Languages and Making Decisions.
• Perform ML Inspired by Human Brain Networks of Neurons.
• Able to Learn without Human Supervision.
• Artificial Neural Networks Adapt and Learn from Vast Amounts of Data.
• Based on Artificial Neural Network

Neural Network

• Pattern Matching through Connection of many Small Functions to Create one very Powerful Function ( Like Neurons in Brains )
• Also Called as Deep Neural Network ( Deep Refers to Number of Hidden Layers Typically )
• Neural Network Connects Simple Nodes | Neurons | Units
• Collection of Such Neurons form a Network ( Recognize Relationships in Data Set )
• Neural Network are Designed to Adapt to Dynamic Input

Components of Neural Network

1. Neurons

• Nodes Receives Data Input and Combine Inputs with Activation Function + Bias

2. Connections and Weights

• Nodes are Connected through Edges
• A Weight is Assigned on each Connection | Edges
• Bias is added at Nodes

3. Activation Functions

• A Function that takes in the Weighted Sum of all the Inputs from Previous Layer + Bias and Generates Output for Next Layer.

1. Sigmoid : form S Shaped Curve ( Range between 0 and 1 )
2. tanh : Hyperbolic Tangent ( Range between -1 and 1 )
3. ReLu : Rectified Linear Unit ( 0 for Negative Input and Same Input for Inputs > 0 ) max(0, x)
4. Leaky ReLu : ( Allow Small Negative Inputs and Returns Same Input for Inputs > 0 ) max(0.1x, x)
5. ELU : Exponential Linear Unit ( Similar to Linear for Inputs > 0 )

4. Learnable Parameters

• Parameters that will be Learned by the Model during the Training ( Weights and Biases )

Artificial Neural Network | ANN

• Simulate the Way the Human Brain Analyzes and Process Information.

• Foundation of AI | Solve Difficult Problems that are Impossible for Human Brain.

• Finding Patterns and Relationships between Data like Human Brains.

• Application

1. Foreign Exchange Trading
2. Forcasting Weather Patterns
3. Speech Recognition

Convolutional Neural Network | CNN

Designed for working with Two Dimensional Image Data

1. Decoding Facial Recognition
2. Analyzing Documents
3. Understanding Climate
4. Image Caption Generation
5. Object Classification

Recurrent Neural Network | RNN

• Feed Forward and Feedback Netwrok
• Better for Sequential Data
• Application

1. Language Translation ( Text Data )
2. Speech Recognition
3. Video Tagging
4. Image Description

CNN ( Faster ) > RNN > ANN

NLP | Natural Language Processing

• Branch of AI
• Computer Learn | Gain | Understand from Human Languages.
• Interaction between Computers and Humans using the Natural Language.
• Objective of NLP is to Read, Interpret, Understand and Make Sense of the Human Languages.
• NLP rely on ML to Derive Meaning from Human Lanuages.

Process :

1. Human Talk to Machine.
2. Machine Capture Audio.
3. Convert Audio to Text.
4. Process Text Data ( Interpret > Convert )
5. Convert Data to Audio.
6. Machine Talk | Reply to Human.

Applications :

1. Google Translate
2. Word Processor | Grammer Check in Microsoft Word.
3. IVR | Interactive Voice Response in Call Centers.
4. Voice Assistant : OK Google, Siri, Cortana and Alexa.

Computer Vision

• Computers Gain High Level Understanding from Digital Images or Videos.
• Works on Pattern Recognition.
• Understand and Automate the Task of Human Visual System.
• Detecting Image | Object and Labeling has Surpassed Humans.
• Faster than Human Reaction and 99% Accuracy
• High Amount of Visual Data ( Images and Videos ) with Processing Capabilities.

Applications :

1. Self Driving Cars
2. Facial Recognition
3. Augmented Reality and Virtual Reality
4. Health Care ( X Ray and MRI Scan )
5. Video Motion Analysis
6. Image Segmentation ( Camera Detects the Multiple Faces in a Group Selfie )
7. Scene Reconstruction ( 3D Model Creation in Architecture )
8. Image Restoration ( Filtering Blur Images and Removing Noise )

Task :

1. Object Classification : Image consist a Cat
2. Object Detection: Classification + Localization : Multiple Objects can be Detected in the Image ( Dog | Cat | Rat )
3. Object Identification : Type of Object in an Image
4. Instant Segmentation : Extract the Images from its Edges | Boundaries.

IoT | Internet of Things

• Connecting Everything in the World to the Internet
• Gives Ability to Send and Receive Information.
• Sea of Data
• Extending Power of Internet beyond Computers and Smartphones to a Whole Range of other Things in the Environments.
• A System of Interconnected Computing Devices, Mechaniacal Devices and Digital Machines | Gadgets.
• Every Machine, Device, Object, Animal and Person is Provided with Unique Identifier.
• Ability to Transfer Data over a Network without Human to Human or Human to Machine Interaction.
• Integrate and Control Everything in the House using Device.

Simple Examples :

1. Search : We can Search | Upload | Download Any Data, Image, Song, Documents, Audio Stored at one Place from Anywhere in this World.
2. Wearables : Fitness Bands | Smart Watches Measures Human Health and Captures Behavioural Patterns and Transfer Data to Database.
3. Health : Wearables consisting of Sensors Monitor Humans Health Patterns and Alert on some Vital Sign | Risk Situation.
4. Traffic Monitoring : Traffics are Visible on Maps, Fastest Route or Alternate Routes.
5. Fleet Management : Connect Fleets of Vehicles with the Drivers and Co Drivers, Fuel, Tyre Pressure, GPS Tracking and Many More.
6. Smart Farming : Quality of Soil, Amount of Water Required, Suitable Crop for Type of Land
7. Hospitality : Electronic Key and Availability of Rooms.
8. Smart Grid and Energy Saving : Grid of Solar Panels, Amount of Energy it Generates, Maintenance

Data Science Process :

1. Obtain Data : Gather Data from Relevant Sources.
2. Scrub : Clean Data to Formats that Machine Understands.
3. Explore : Find Significant Patterns and Trends using Statistical Methods.
4. Model : Construct Models to Predict | Forecast | Estimate.

Back to Questions