AI textbooks define AI as “The Study and Design of Intelligent
Agents” where an intelligent agent is a system that perceives its
environment and takes actions that maximize the chances of its success.
The Capability of Machine to simulate and
Imitate Intelligent Human Behavior is Artificial Intelligence.
Application of AI: Some examples borrowed from Course Era
- Movie/ Songs recommendation
- Fraud Detection – credit card purchases
- Smart home devices
- Facebook – tagging friends
- Video Games
- Virtual Personal Assistants: Siri, Cortana
The brain behind AI is the Machine Learning , Deep Learning and Natural Language Processing (NLP).
Some scientist classify NLP as Deep Learning and some scientist classify NLP as Machine Learning.
Important Components of Deep Learning and Machine Learning.
(NLP will be covered in the following post)
Lets start with Data:
As seen below the type of Statistical Methods keep changing with DATA COMPLEXITY
As the data complexity increases - there is a shift from Traditional Statistical Methods to Small Neural Network to Medium Neural Network and then Large Neural Network
Algorithms : They can be Machine Learning or Deep Learning Algorithms
What is Machine Learning:
•
Algorithms that do the learning
without human intervention.
•
Supervised Learning: Learning with a labeled training set (cases are
labelled as – spam / no spam , fraud /no fraud already available in the training data-set )
•
Unsupervised Learning : Discover patterns in unlabeled data (labeling of cases not available)
•
Optimize parameters on training data using Gradient Descent
Computation
•
Goal of Machine learning –minimization of error
(predicted – actual).
Machine Learning Algorithms:
Linear Regression
Logistic Regression
Cluster Analysis
Support Vector Machine
Shallow Neural Network
Principal Component Analysis
Recommend-er System
Anomaly Detection /Outlier
detection
Now let us take the example of shallow neural network .
The below neural network with one hidden layer is shallow neural network or simple neural network that falls under Machine Learning
Why is Deep Learning Taking of?
Need for Deep Learning arise when the data complexity increases.
Complex Neural Network falls under Deep learning
A complex neural network is Deep Learning as shown below
A deep neural network with many hidden layers falls under DEEP LEARNING
Applications of Deep Learning
Computation :
Mathematical computation is required to derive co-efficient / parameter estimates / Weights for the Model.
Co-efficient is derived using Differential Calculus , the technique is known as - "Gradient Descent"
Both the Machine Learning and Deep Learning Algorithm uses "Gradient Descent" for Computation of co-efficient.
All algorithms needs computation of parameter estimates which is done using a differential calculus technique called as - "Gradient Descent"
So for now , lets conclude the post .
I will cover the details in the coming post.
Comments
Post a Comment