Data Scientist @ Siemens Advanta Consulting | Master’s in Artificial Intelligence, NUS | Love to work on Frontier Technologies | Singapore | http://msundarv.com
Image for post
Image for post
Self-Supervised Representation Learning based Upstream Models for Acoustic Data — wav2vec [1], Mockingjay [4], Audio ALBERT [5], vq-wav2vec [3], CPC[6]

People following Natural Language Processing (NLP) research closely will know that latterly introduced language models like BERT [7], GPT are radically changing the NLP field. In NLP, unsupervised pre-training of language models improved many tasks such as text classification, semantic textual similarity, machine translation. Have you ever wondered how these unsupervised language model pre-training can help sequential time series data apart from the free text?

Recently I got an opportunity to work on acoustic data where we have to recognize the running component, say motor or compressor, of a machine based on the recorded sound data. Since there were no labeled data for trying out supervised learning, I had to look for alternatives. …


Image for post
Image for post
Weight Poisoning Attacks on Pre-trained Models — Photo by Author based on [1]

I am confident that most of us are familiar with backdoor attacks when it comes to computer security, but have you ever imagined that your deep learning models are at risk because of similar attacks. As Machine Learning Practitioners, we should be aware of theses attacks to secure our models.

Transfer learning is where knowledge learned by a previously trained model is transferred for training a new model. Transfer learning plays a significant role when the target problem has fewer data and is the primary reason behind the commercial success of machine learning. …


Image for post
Image for post

Want to go beyond English and use the real power of Natural Language Processing (NLP) to serve the second-most populous country in the world? Everyone knows India is a very diverse country and a home for lots of languages but do you know India speaks 780 languages [1]. It’s time to go beyond English when it comes to NLP. This article is for those who somewhat know about NLP and want to start using it for Indian languages.

Language Models

Before going into the topic, we will glance at some essential concepts and recent accomplishments in NLP. NLP helps computers in understanding human language. Text Classification, Information Extraction, Semantic Parsing, Question Answering, Text Summarization, Machine Translation, and Conversational Agents are some applications of NLP. …


Image for post
Image for post

I am sure most of you would have come across Named Entity Recognition (NER). NER is a fundamental Natural Language Processing (NLP) task and has a wide range of use cases. This article is not about NER but about an NLP task that is closely related to NER.

Do you know what is Named Entity Linking (NEL)? How does it help in Information Extraction, Semantic Web, and many other tasks? If not, then don’t worry. This article will answer those questions along with a basic implementation of NEL.

Before looking into NEL, we will first understand information extraction. …


Image for post
Image for post

Recently I got an opportunity to work on Survival Analysis. Like any other project, I got excited and started to explore more about Survival Analysis. As per wiki,

“Survival analysis is a branch of statistics for analyzing the expected duration of time until one or more events happen, such as death in biological organisms and failure in mechanical systems.”

In short, it is a time to event analysis that focuses on the time at which the event of interest occurs. The event can be death, sensor failure, or occurrence of a disease, etc. …

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store