Neural Networks

Another perspective on coffee — My professional career has been a mashup of data science, computer vision, and machine learning. When I started to apply these skills to coffee, I found a multi-variable process that was challenging and interesting. …

Preventing Abuse of LLMs’ Alignment Deficit by Injection Neutralization (PALADIN) Out of all the Large Language Models (LLMs) currently out in the open, I’ve found Claude+ to be by far the safest and most harmless one. The team at Anthropic, a cutting-edge AI startup valued at $4B+, has done an absolutely brilliant job taking AI Safety to the next level…

Optimizers are an essential component of neural networks that are responsible for updating the weights of the neural network during the training process. They are algorithms that help in finding the optimal set of weights that minimize the loss function. A loss function measures the difference between the predicted and…

Competition between transformers and convolutional neural networks in computer vision has provoked a rapid evolution of both since the beginning of 2020s. In the last few years, community discovered not only that attention-based architectures are a good fit for CV tasks, but also that convolutional networks still have a great…

This is the first post in the series of upcoming posts about Generative Deep Learning Models driving the whole new Generative AI revolution in 2023 Machine learning models can be broadly divided into discriminative or generative models. To understand the idea behind both these models, Lets take a toy dataset…

As a data scientist for my project in college, I was assigned the challenging task of building an image classification model to identify specific features from product images and generate tags. …

solving inverse differential equation problems — (1) Introduction: What is physics-informed? Many relationships in physics, biology, chemistry, economics, engineering, etc., are defined by differential equations. (Check here for an extensive list.) In general, a differential equation (DE) describes how variables are affected by the rate of change of other variables. For instance, a DE explains how…

This will be a short post explaining what does `torch.nn.init.orthogonal_()` do and how to use it. Let’s define a simple layer and look at its weights before and after applying the function. layer = nn.Linear(2,2) print(layer.weight) # And let's look at its eigenvalues print(torch.abs(torch.linalg.eig(layer.weight)[0]))