dl201's Introduction

DL201

Deep learning 201

• Week 1
    ◦ Concepts
        ▪ Tensors
        ▪ Indexing
        ▪ Broadcasting/vectorization
    ◦ Skills
        ▪ Choosing a Dataset
            • Key Questions
                ◦ What's problem we want to solve?
                ◦ What data sources exist?
                ◦ Where can we store this data?
            • Metadata Analysis
                ◦ Sample Size
                ◦ Class Balance
                ◦ What features does it contain?
        ▪ Loading Tabular Data + EDA
            • Basic Data Wrangling
                ◦ Change Column Names
                ◦ NaN Values
                ◦ Converting text to int 
                ◦ Outliars
            • EDA (Summary Statistics)
                ◦ Sample Size
                ◦ Features
                ◦ Missing Values? What to do?
                ◦ Outliers 
                ◦ Ask Domain Expert?
        ▪ Loading image data + EDA
            • Data Wrangling
                ◦ Directory Structure
                ◦ Label Extraction
                ◦ Feature Extraction
            • EDA
                ◦ Sample Size
                ◦ Class Balance 
                ◦ Features present
                ◦ HITL Tools 
        ▪ Loading NLP data + EDA 
            • Data Wrangling
            • EDA Stats
• Week 2
    ◦ Concepts
        ▪ Continuous Variables 
            • Pre-processing
                ◦ Make it easier for computer to see something
            • Transformation 
                ◦ Normalization
        ▪ Categorical Variables
            • Label Encoding
            • One hot encoding
            • Numeric encoding
        ▪ Embeddings
            • Word Embeddings
            • Category Embeddings
    ◦ Skills
        ▪ CV
            • Pixel Transforms
                ◦ Filters
                    ▪ Remove noise
                    ▪ Blurring Images
                ◦ Edge Detection
                ◦ Contrast Adjustment
                ◦ Equalization (zone adjustment)
        ▪ TAB
            • Preprocessing 
                ◦ Indicating Features
                ◦ Discretisation 
                ◦ Interaction Features
                ◦ Feature Decomposition
            • Transforms
                ◦ One-Hot Encoding
                ◦ Category Embeddings
        ▪ NLP
            • Lemmatization
            • Simple NLP pre-proccessing

Recommend Projects