Sample and Event
⭐ Sample vs. Event Think of probability as a story with two levels: 🎯 Sample Space (S) The sample space is the complete list of everything that could happen in an experiment. Examples Key idea…
AI
Downsampling Techniques for Custom Batches in training Machine Learning Models
Downsampling techniques can be used to create custom batches for training a machine learning model. These custom batches can be tailored to specific needs, such as improving training efficiency, handling imbalanced data, or focusing on…
downsampling for hyperparameter tuning
Downsampling for hyperparameter tuning reduces the dataset size to speed up model training and experimentation while preserving key data characteristics. Here’s a concise overview: Why Downsample for Hyperparameter Tuning? Key Considerations Practical Steps Pitfalls to…
How to reduce catastrophic forgetting when fine-tuning neural networks
Catastrophic forgetting occurs when a neural network “overwrites” what it learned in a previous task while training on a new one. This happens because the weights optimized for the first task are changed to minimize…
Textbooks Are All You Need
The central idea of the paper “Textbooks Are All You Need” is that data quality is significantly more important than data quantity or model size when training large language models (LLMs), particularly for code generation.…
The “Speed” Myth of sloths
Baby sloths are born with their eyes open, all their teeth, and the ability to climb immediately.
Sloth’s Extreme Digestion & Bathroom Habits
Extreme Digestion & Bathroom Habits A Walking Ecosystem
Geckos’ amazing feet
Geckos are arguably the most charismatic and bizarre family in the reptile world. They are the only lizards that can vocalize, many can climb glass, and some have superpowers that seem pulled straight from a…
Komodo dragons with stinky breath
Komodo dragons are the closest thing we have to real-life dinosaurs. They are the heaviest lizards on Earth, but their size is just the beginning of what makes them interesting. Here are the most fascinating…
What’s Cache Augmented Generation (CAG)
Cache Augmented Generation (CAG) is an architecture for Large Language Models (LLMs) that removes the need for real-time data retrieval by pre-loading a knowledge base directly into the model’s active memory. In practical terms, while…
Relationship Between MAE, MSE, RMSE
📐 Definitions (for clarity) Let errors be . MAE MSE RMSE 📐 Relationship Between and Let the errors be Then: Mean Absolute Error (MAE) Mean Squared Error (MSE) Key Relationship 1. Jensen’s Inequality gives: Why?…
Similarity Metrics for MR Image-to-Image Translation
The major conclusion of the paper Similarity Metrics for MR Image-to-Image Translation is that relying on the most commonly used metrics, specifically SSIM and PSNR, is insufficient for validating Magnetic Resonance (MR) image-to-image translation models…
Polynomial regression in Python
Polynomial regression is a form of regression analysis where the relationship between the independent variable and the dependent variable is modeled as an degree polynomial. Polynomial regression fits a nonlinear relationship between the value of…
Why is contrastive learning?
Contrastive learning is a technique used in machine learning, particularly in the field of self-supervised and unsupervised learning. It focuses on learning to distinguish between similar and dissimilar pairs of data points by contrasting them…
Choosing a Boyfriend Explained: Dating as a Maximum Likelihood Problem with the EM Algorithm
Choosing a boyfriend as a maximum likelihood problem can be framed as an exercise in probabilistic decision-making, where the goal is to maximize the likelihood of selecting a partner who best fits your desired criteria…
An overview of explainable AI techniques
Explainable AI (XAI) techniques are methods and processes used to make AI models and their predictions understandable to humans. These techniques are critical for building trust, ensuring ethical use, and meeting regulatory requirements in AI…
Common distance measures in machine learning and their properties
Common distance measures in machine learning, their formulas, use cases, and detailed properties: 1. Euclidean Distance 2. Manhattan Distance (L1 Norm) 3. Minkowski Distance 4. Cosine Similarity 5. Hamming Distance 6. Jaccard Distance 7. Mahalanobis…
Attention Transfer for Convolutional Neural Networks
The paper “Paying more attention to attention: improving the performance of convolutional neural networks via attention transfer” by Sergey Zagoruyko, Nikos Komodakis Université proposed a novel training methodology called attention transfer to improve the performance of convolutional neural…
Physics-Informed Neural Networks for Brain Molecular Transport from MRI
The method described in the paper “Investigating molecular transport in the human brain from MRI with physics‑informed neural networks” by Bastian Zapf, Johannes Haubner, Miroslav Kuchta, Geir Ringstad, Per Kristian Eide, Kent‑Andre Mardal centres on…
brain datasets that have missing mri sequences
Finding datasets that naturally feature missing sequences (where patients were simply not scanned with the full protocol) is rare because public repositories usually curate data to ensure completeness. However, several high-profile datasets are specifically designed…
EM algorithm
The Expectation-Maximization (EM) algorithm is an iterative approach to estimate the parameters of probabilistic models, such as a Gaussian (normal) distribution, when the data is incomplete or has missing values. It alternates between two steps:…
ImputeFormer: Low Rankness-Induced Transformers forGeneralizable Spatiotemporal Imputation
The paper “ImputeFormer: Low Rankness-Induced Transformers for Generalizable Spatiotemporal Imputation,” from the KDD ’24 conference, addresses the pervasive issue of missing data in spatiotemporal datasets, particularly contrasting traditional low-rank models with modern deep learning methods like…
Classification via Label Imputation and Imputation Using Labels
The paper Imputation Using Training Labels and Classification via Label Imputation introduces two novel machine learning algorithms designed to efficiently handle missing values, a common issue in practical datasets. The first approach, Classification Based on…
Feature-Based Knowledge Distillation in Pytorch on MNIST
In this post, we will talk about feature-based knowledge distillation. One of the pioneering paper is “FitNets: Hints for Thin Deep Nets” by Adriana Romero, Nicolas Ballas, Samira Ebrahimi Kahou, Antoine Chassang,Carlo Gatta & Yoshua…
Example of Knowledge Distillation (KD) using PyTorch on the MNIST dataset
This example demonstrates Knowledge Distillation, a technique where a small “student” model is trained to mimic a larger, pre-trained “teacher” model. Let’s have a brief introduction to Knowledge Distillation first. 🎓 What is Knowledge Distillation?…
relation between knowledge distill and perceptual loss
Knowledge distillation and perceptual loss are distinct concepts in machine learning, but they can be used together effectively, especially in computer vision tasks. Here’s the simple breakdown: 🧠 What is Knowledge Distillation? Knowledge Distillation is…
Using Rust with Visual Studio Code
Visual Studio Code is the most popular and recommended code editor for Rust development. The experience is excellent, but it requires one key extension to make it work. How to Set It Up (2 Steps)…
RUST for AI software development
What is RUST? Rust is a modern systems programming language focused on three core goals: performance, memory safety, and concurrency. Think of it as having the raw speed and low-level control of languages like C…
Difference between MRI and fMRI
The simplest way to think about it is: Think of an MRI as a high-resolution photograph of the brain, showing you exactly what it looks like. An fMRI is more like a video or a…
MONAI Framework for Advanced Brain Image Analysis
The application of artificial intelligence (AI) in healthcare, particularly in medical imaging, holds immense promise for improving the detection, diagnosis, and treatment of human diseases. However, the translation of AI models from research prototypes to…