Variational Inference Variational Inference (VI) is a strategy to approximate difficult-to-compute probability densities (Blei et al., 2017). Being an alternative method to the Markov chain Monte Carlo sampling (MCMC), it is widely used in Bayesian models to compute the approximate posterior probability densities. Its applications encompass computer vision, computational neuroscience and large…

ArXiv Enhanced is a browser extension which can be used to add comments, tags and other details for an arXiv paper and store them in a notion database. This article covers tips on building the chrome plugin, building with webpack, and touches upon the notion API. In a later article I will discuss other features of the extension — adding notes to pdf, highlighting and OAuth. — TL;DR Implementing a Chrome plugin that connects with a Notion database

A while ago, we formulated a variant of Sudoku called the “bridge” Sudoku. We were curious on how it compared to solving a typical Sudoku in terms of difficulty. We encode bridge Sudoku based on the minimal encoding scheme. The statistics given by the minisat solver are used to comment on the difficulty of solving the Sudoku. We found that considering the above statistics, it is easier to solve a bridge Sudoku against solving the corresponding standard Sudokus.

Variational Auto-Encoders are known to be susceptible to posterior collapse, where the approximate posterior collapses to the prior distribution due to the vanishing of the KL term (Bowman et al., 2015 [1]). This forces the approximate posterior to be independent of the data, i.e., q(z|x) = q(z) = p(z), where…

In this project, we aim to model the inflected word forms. We study two generative models namely, Character VAE, a single sequence variational auto encoder with only continuous variables for sequential data and, MSVAE (Multi-Space Variational AutoEncoder), a generative model with both continuous and discrete latent variables. The relevant experiments…

This is an excerpt from my master thesis titled: “Semi-supervised morphological reinflection using rectified random variables” In this story we describe the stretch-and-rectify principle applied to the Kumaraswamy distribution [1]. This technique was proposed by Louizos et al., 2017 [2] who rectified samples from a Gumbel-sigmoid distribution.

Building a centralized system for covid19 resource leads, co-verified by the people: coverify.in — During the second wave in India, there was a spike in the number of covid19 infections, with number of cases exceeding 400,000 per day at the peak (Source: Google). The sudden increase tested the various systems in place — hospital capacity, access to valuable resources, mental support, etc. Having heard…

IPC with Named Pipes Recently, I came across a use-case which involved running a PyTorch model for inference in real-time, along with a Node.js server which handles I/O. After an initial failed attempt of using torch.js, we decided to go with Tensorflow.js (tfjs). However, the layers model is a synchronous…

This is an excerpt from my master thesis titled: “Semi-supervised morphological reinflection using rectified random variables” Introduction Languages use suffixes and prefixes to convey context, stress, intonation, and grammatical meaning (like subject-verb agreement). Such suffixes and prefixes form a more general class of entities which are the meaningful sub-parts of a…