Machine Learning

Machine Learning vs AI

• AI aims to help machine carry out task, whereas Machine Learning aims to identify patterns in the data

• AI are broadly categorized into 2 buckets - generalized and applied.

• Applied AI is focussed to solve a specific task - for example, wealth management, autonomous car

• Generalized AI - "one ring that that will rule them all" ... well, we are not there yet!

• What factors are pushing forward the development in Machine Learning today?

  • Realization that machines can instructed to learn by themselves using data

  • Abundance of available data that can fodder the machines

Neural Network

A computer system designed to work like human brain to classify information. It is used to solve some of these tough

• Infer the author's sentiment behind piece of text

• Whether a piece is music is likely to make one happy or sad

• Infer the meaning of natural language in the form of text or audio

AI And Machine Learning Use Cases

• Data Security - identify the pattern of data access of malware to detect breaches

• Personal Security - analyze security security screenings at airports, large gathering to identify threats as much faster rate and higher accuracy than human screeners

• Financial Trading - find which stocks will rise and which will fall

• Healthcare - understand the risk factors of diseases in large population, media diagnostics

• Marketing - personalized ad targeting on web properties

• Fraud Detection - identity and prevent fraudulent transactions in e-commerce and banking

• Recommendation - product / service recommendation

• Search - refine searches based on the context, search history and user profile

• Natural Language Processing - simplify the essential meaning of a text for fast consumption

Github Repo with code solutions

• Machine learning with Scikit Learning, and Spark ML https://github.com/abulbasar/machine-learning

• Neural Networks with Tensorflow, Keras etc. https://github.com/abulbasar/neural-networks

In the above repositories you can see code examples for the following case studies

• Estimation for insurance premium based on the customer demographic information

• House price prediction problems based on Kaggle competition data

• Power demand forecasting based

• Stock price forecasting using time series analysis

• Credit risk assessment

• Credit card transaction fraud detection

• Predict customer participation of bank marketing campaign

• Ad-click prediction using web analytics

• Customer churn prediction for tele comm customers

• Segmentation of retail customers

• OCR - image classification

• Image classification using cutting edge deep learning for high accuracy

• Detection of DOS attack using the netflow data

• Sentiment analysis

• Movie recommendation