Radim Urban | Blog

SVHN Image and Label Synthesis with Conditional DCGAN

We implement a conditional Deep Convolutional Generative Adversarial Network (DCGAN) sampling high-quality Street View House Numbers (SVHN). Apr 19, 2024 - EESTec AI Marathon (3rd place)

Using LSTM RNNs to Predict Sepsis from Clinical Data

Training a Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) model on raw time-series clinical data for early sepsis detection. Mar 30, 2024

Intuitive Understanding & In-Code Implementation of Dropout

Dropout helps to prevent overfitting and encourages the network to be more robust by preventing reliance on specific neurons. Nov 26, 2023

Prompting LLM to get protected information

Observing and passing Gandalf The White by prompting an LLM to get protected information. Nov 15, 2023

Principal Component Analysis - Derived and Coded

Principal Component Analysis (PCA) is a method to get low dimensional representation of a high-dimensional dataset without losing the important features. Jul 7, 2023