Difference between AI, ML, and DL
Artificial Intelligence or AI works without human interference.
Machine Learning or ML is a subset of AI where data is most important.
Deep Learning mainly comprises Neural Networks which mimic the brain.
Data Scientist is a person who works on stats tools to analyze, preprocess, predict and forecast.
AI applications
Artificial Intelligence has various applications in the modern world like Netflix, Self-Driving cars, Amazon, Youtube.
Types of ML techniques
Supervised Machine Learning
Training data has output along with the input and hence train according to the output provided to the given input.
- Regression — Output consists of continuous values
- Classification — Output consists of a fixed number of categories
- Binary Classifier — Classifies in mainly two categories (0 or 1).
- Multiple Classifier — Labels are mutually exclusive.
- Multilabel Classifier — An object can be categorized in multiple labels.
Note- Multiclass has a single output and multiple categories whereas multilabel have multiple outputs and multiple categories.
Forecasting is a regression problem. Eg- Time series analysis.
Unsupervised Machine Learning
Training data doesn’t have output along with the input. It recognizes patterns in the data and tries to classify them based on that. These are three types of categories:
- Clustering
- Segmentation
- Reduced Dimensions
Semi-supervised Machine Learning
It is trained on labeled data and then it recommends the solution. Based on interaction, the model gives new recommendations.
Eg- Newborn babies don’t know anything initially. They walk and fall. Based on their experience they learn where to walk and where not.
Deep Learning
- Artificial Neural Network- It works on tabular data
- Convolution Neural network — It works on images and videos.
- Image Classification
- Object Detection
- Object Segmentation
- Tracking
- GAN
- Recurrent Neural Network — In this, the input data is text, time series, sequential data.
- Natural Processing Language — It works on text data mainly.
