Machine Learning Basics
Anuj Mahajan
Assistant Professor
School of CSE
Shri Mata Vaishno Devi University
Artificial Intelligence(AI), Machine
Learning (ML), Deep Learning (DL)
• AI is an umbrella term that covers everything related to
making machines smarter.
• ML is a subset of AI, performance of ML algorithm improves
over a period of time as its experience increases.
• DL is a subset of ML. DL uses Deep Neural Network to improve
its accuracy. Its performance is far better than general ML
algorithms, if Deep Neural Network is trained on a very large
training set.
Three broad categories Of AI
• Artificial Narrow Intelligence (ANI) is Weak AI:
Such systems are good at performing a
particular task e.g. the search algorithms in AI.
• Artificial General Intelligence (AGI) is Strong
AI. These are autonomous systems (agents)
with Human like Intelligence
• Artificial Super Intelligence (ASI) is Strong AI..
These are autonomous systems (agents) with
Intelligence beyond human beings
Machine Learning
Courtesy: Andrew NG’s ML course (CourseEra)
Machine Learning
Courtesy: Andrew NG’s ML course (CourseEra)
Machine Learning
Courtesy: Andrew NG’s ML course (CourseEra)
ML model
Courtesy: [Link]
Neural Network
Courtesy [Link]
Deep Neural Network
Courtesy: [Link]
Deep Neural Network Layers
Courtesy: [Link]
Deep Learning Success
Courtesy: [Link]
• Near-human-level image classification
• Near-human-level speech recognition
• Near-human-level handwriting transcription
• Improved machine translation
• Improved text-to-speech conversion
• Digital assistants such as Google Assistant and Amazon
Alexa
• Near-human-level autonomous driving Improved.
• Asd targeting, as used by Google, Baidu, and Bing
• Improved search results on the web
• Superhuman Go-playing
Machine Learning
Courtesy: Andrew NG’s ML course (CourseEra)
Supervised Learning- Regression
Courtesy: Andrew NG’s ML course (CourseEra)
Supervised Learning (Classification)
Courtesy: Andrew NG’s ML course (CourseEra)
Example
Courtesy: Andrew NG’s ML course (CourseEra)
Ensemble Method–Random Forest
Classifier
Courtesy: [Link]
general-concept-
Unsupervised Learning (Clustering)
Courtesy: Andrew NG’s ML course (CourseEra)
Clustering New Articles
Courtesy: Andrew NG’s ML course (CourseEra)
Genetics
Courtesy: Andrew NG’s ML course (CourseEra)
Other Applications of Clustering
Courtesy: Andrew NG’s ML course (CourseEra)
Example
Courtesy: Andrew NG’s ML course (CourseEra)
Dimensionality Reduction
(Unsupervised Learning )
Courtesy: Andrew NG’s ML course (CourseEra)
Dimensionality Reduction (e.g. PCA algorithm)
Courtesy: Andrew NG’s ML course (CourseEra)
Association Rule Mining
(Unsupervised)
Courtesy: [Link]
Anomaly Detection
Courtesy: Andrew NG’s ML course (CourseEra)
Reinforcement Learning
Courtesy: Wikipedia
Recommender Systems
• NetFlix , Amazon, Facebook etc. most sites use
these to recommend you products.
• In the backend a ML system is working.
References
1. Andrew NG’s course on Machine Learning at
CourseEra
2. [Link]
3. Wikipedia
4. [Link]
5. [Link]
6. [Link]
classifier-random-forest-classifier-general-
concept-