Deep learning
Neural networks from the ground up: layers, optimisers, CNNs, RNNs, GANs and training tricks.
Page 2 of 10
- Regression and Classification Loss Functions8 min
Mean Bias Error (MBE) captures the average bias in predictions. It's rarely used for training as positive and negative errors can cancel each other out...
- Regression and Classification Loss Functions Cheat Sheet10 min
Mean Bias Error provides insights into the systematic over or underestimation in regression models by calculating the average difference between predicted...
- Response:8 min
Neural networks transform input data through layers of interconnected nodes, with activation functions determining the output of each neuron. These...
- Sigmoid Activation Function in Python7 min
The sigmoid activation function is a fundamental component in neural networks and machine learning. It maps input values to an output range between 0 and 1...
- Softmax Activation Function in Neural Networks with Python10 min
The Softmax activation function is a crucial component in neural networks, particularly for multi-class classification problems. It transforms a vector of...
- Softmax and Categorical Cross-Entropy Using Python10 min
Softmax and Categorical Cross-Entropy are fundamental concepts in machine learning, particularly in classification tasks. Softmax transforms raw model...
- Softmax and Sharp Decision-Making in AI Systems7 min
The softmax function is a crucial component in modern AI systems, enabling differentiable query-key lookups and sharp decision-making. This presentation...
- Softmax Function Explained Transforming Logits to Probabilities7 min
The Softmax function transforms a vector of real numbers into a probability distribution, ensuring all values are between 0 and 1 and sum to 1. It's commonly...
- Softmax in Machine Learning with Python9 min
Softmax is a crucial function in machine learning, particularly in multi-class classification problems. It transforms a vector of real numbers into a...
- Stable Minima in Univariate ReLU Networks9 min
Stable minima play a crucial role in the generalization capabilities of neural networks. In this presentation, we'll explore how stable minima cannot overfit...
- Types of Activation Functions10 min
The sigmoid function transforms input values into outputs between 0 and 1, making it useful for binary classification problems and probability outputs. It...
- Understanding Activation Functions in Neural Networks10 min
Neural networks require non-linear activation functions to model complex patterns. We'll implement the classical sigmoid and tanh functions from scratch...
- Understanding AI Through ReLU Activation Function in Python7 min
Introduction to Activation Functions in Neural Networks
- Understanding the Role of ReLU Activation in Neural Networks5 min
The Rectified Linear Unit (ReLU) activation function transforms input by outputting zero for negative values and maintaining positive values unchanged. This...
- Why ReLU Function is Not Differentiable at x=07 min
The Rectified Linear Unit (ReLU) function is a crucial activation function in neural networks. It's defined as f(x) = max(0, x), which means it outputs x if...
- A Simple Neural Network Module for Relational Reasoning11 min
Relational reasoning is a fundamental aspect of human intelligence, allowing us to understand and manipulate relationships between entities. This slideshow...
- Anomaly Detection in Graph-Based Data Using Latent Space Diffusion Models9 min
Anomaly detection in graph-based data is a crucial task in various domains, including social network analysis, fraud detection, and cybersecurity. This...
- Artificial Neurons for Modeling Local Joint Distributions in Python:10 min
Artificial neurons are computational units inspired by biological neurons, forming the foundation of artificial neural networks. In this context, we'll...
- Autoencoders for Manifold Dimension Discovery in Python10 min
Autoencoders are neural networks designed to learn efficient data representations in an unsupervised manner. They can be used to discover the intrinsic...
- Boundary Integrated Neural Networks vs Physics Informed Neural Networks14 min
Boundary Integrated Neural Networks (BINNs) and Physics Informed Neural Networks (PINNs) are advanced machine learning techniques used to solve differential...
- Categorizing Research Papers with Graph Neural Networks in Python9 min
Graph Neural Networks (GNNs) have emerged as a powerful tool for analyzing and categorizing research papers. By representing papers and their relationships...
- Comparing Physics Informed Neural Networks (PINN) and Finite Element Method (FEM) in Python11 min
Physics Informed Neural Networks (PINN) and Finite Element Method (FEM) are two powerful approaches for solving complex physical problems. While FEM has been...
- Compressing ML Models with Knowledge Distillation12 min
Knowledge distillation is a technique used to compress large, complex machine learning models into smaller, more efficient ones while preserving their...
- Constructing MaxOut Neural Networks in Python12 min
MaxOut networks are a type of neural network architecture introduced by Ian Goodfellow et al. in 2013. They are designed to improve the performance of deep...
- Credit Card Fraud Detection with Autoencoders in Python7 min
Credit card fraud is a significant problem for financial institutions and consumers alike. Machine learning techniques, particularly autoencoders, can be...
- Deep Learning Autoencoders in Python9 min
Autoencoders are neural networks designed to learn efficient data representations in an unsupervised manner. They compress input data into a...
- Efficient Knowledge Distillation Techniques12 min
Knowledge distillation is a model compression technique where a smaller student model learns to mimic the behavior of a larger teacher model. The process...
- Exploring AutoEncoders in TinyML Foundations, Training, and Applications13 min
AutoEncoders are neural networks designed to learn efficient data representations in an unsupervised manner. In the context of TinyML, they play a crucial...
- Graph Neural Networks in Python13 min
Graph Neural Networks are a class of deep learning models designed to work with graph-structured data. They can capture complex relationships between...
- Graphical Neural Networks in Python:10 min
Graphical Neural Networks (GNNs) are a powerful class of deep learning models designed to work with graph-structured data. They leverage the relationships...