Software Engineering Module 3: Software Automation

Key Characteristics of Machine Learning (ML)

Learning from data, Prediction and decision-making, Automation

DevOps

Set of practices that combines software development (Dev) and IT operations (Ops)

Functions of DevOps

Predictive Analytics in CI/CD Pipelines, Intelligent Resource Management, Automated Incident Management

MLOps

Automated process of designing, training and deploying machine learning models

MLOps Design Stage

Defining business problem, refactoring business problem into machine learning problem, defining success metrics, researching available data

MLOps Model Development Stage

Data wrangling, feature engineering, model training, model testing and validation

MLOps Operations Stage

Model deployment, supporting operations/use, monitoring model performance

Robotic Process Automation (RPA)

Uses software robots or “bots” to automate repetitive, rule-based tasks traditionally performed by humans

ML in RPA

Handling Unstructured Data, Intelligent Decision-Making

Business Process Automation (BPA)

Use of technology to automate complex business processes, aiming to increase efficiency, reduce human error, and improve overall productivity

ML in BPA

Automating Decision-Making, Handling Unstructured Data in Business Workflows

Artificial Intelligence (AI) vs. Machine Learning (ML)

AI: making decisions, understanding languages, recognising objects, playing games
ML: (specific method within AI) learn from data, identify patterns, make decisions with minimal human intervention

ML Training Models

Supervised, Unsupervised, Semi-Supervised, Reinforcement

Supervised Learning Use Case

Predictive or Classification

Unsupervised Learning Use Case

Pattern Finding, Clustering

Semi-Supervised Learning Use Case

Combination of Supervised and Unsupervised

Reinforcement Learning Use Case

Decision-Making Over Time

Convolutional Neural Networks (CNNs)

Used to identify patterns within images

Neural Networks

Inspired by the biological neural networks in the human brain

Neural Network Structure

Input Layer, Hidden Layers (activation functions), Output Layer

Feedforward Neural Networks (FNN)

Data moves in one direction, from input to output, without cycles or feedback loops

Recurrent Neural Networks (RNN)

Designed for sequence data, have feedback connections that allow information to be passed from one time step to the next

Deep Neural Networks (DNN)

Multiple hidden layers to learn more abstract and complex features from the data

Generative Adversarial Networks (GANs)

Generator (creates fake data) and discriminator (evaluates whether data is real or fake) are trained together

Types of Machine Learning Algorithms

Linear Regression, Logistic Regression, K-Nearest Neighbour (KNN)

Key Concepts of Neural Networks

Neurons and Layers, Feedforward Process, Activation Function, Backpropagation and Gradient Descent

Impact of Automation

Safety of Workers, People with Disability, Skills Required for Development, Efficiency Waste and Environment, Economy and Wealth

Human Behaviour Influences

Psychological Responses, Acute Stress Response, Cultural Protocols, Belief Systems

AI Bias Types

Human Bias, Dataset Source Bias