From Jaccard to OpenAI, implement the best NLP algorithm for your semantic textual similarity projects — 🎬 Introduction Natural Language Processing (NLP) has tremendous real-world applications in information extraction, natural language understanding, and natural language generation. Comparing the similarity between natural language texts is essential to many information extraction applications such as Google search, Spotify’s Podcast search, Home Depot’s product search, etc. The semantic textual similarity (STS) problem…

Blazing Fast, Highly Scalable Text-to-Image Search with CLIP embeddings and Milvus — Introduction I introduced three simple methods to convert images into embeddings for similarity search applications using state-of-the-art neural networks in my previous post. In this post, let's discuss how we can use those embeddings together with Milvus, one of the most popular open-source vector search databases, to create a production scale…

Long live ANNs for their whopping 380X speedup over sklearn’s KNN while delivering 99.3% similar results. — We’re living through an extinction-level event. No, not COVID19. I’m talking about the demise of the popular KNN algorithm that is taught in pretty much every Data Science course! Read on to find out what’s replacing this staple in every Data Scientists' toolkit. KNN Background Finding “K” similar items to any given…

Easiest ways to perform image-to-image and text-to-image similarity search — A whopping 90% of data created since the dawn of human civilization was produced in the past two years! The rate of data creation continues to increase with the proliferation of digital technologies such as social media and the internet of things (IoT) together with ever-faster wireless communication technologies such…

Use domain knowledge to generate large labeled datasets with state-of-the-art NLP Weak Supervision. — 🚧 The challenge of contemporary Machine Learning One of the major bottlenecks for developing modern machine learning (ML) models in real-world applications is the need for substantial amounts of manually-labeled training data [Paper]. For example, the ImageNet dataset consists of over 14Million manually labeled images of various real-world objects. …

Find out a name’s likely gender using Natural Language Processing in Tensorflow, Plotly Dash, and Heroku. — Choosing a name for your child is one of the most stressful decisions you’ll have to make as a new parent. Especially for a data-driven guy like me, having to decide on a name without any prior data about my child’s character and preferences is a nightmare come true! Since…

Setup & Run SQL in Google Colab with just 2 helper functions! — Coding tests are pretty much standard in Data Science interview processes these days. As a Data Science hiring manager, I find a 20–30 min live coding test with some prepared tasks to be effective at identifying candidates who would be successful in the roles that I typically hire for. Google…

~80% of what you need to know about logging in under 5 mins — There comes a time in every production Data Science project when the code base has become complex, and a refactor is necessary to maintain your sanity. Perhaps you want to abstract out commonly used code into Python modules with classes and functions so that it can be reused with a…

What you can do when someone blatantly plagiarises your work on Medium — Publishing Data Science stories on Medium is hard work. It takes weeks (even months) to research interesting topics, architect code in the simplest way possible, and weave it all into an engaging story. …