Zero-Shot Learning in Modern NLP

By Joe Davison Blog - 2020-05-29

Description

State-of-the-art NLP models for text classification without annotated data

Summary

Natural language processing is a very exciting field right now.
However, models of this size remain impractical for real-world use.
If we were to use word vectors as our label representations, however, we would need annotated data to learn an alignment between the S-BERT sequence representations and the word2vec label representations.
In the most recent version of their paper, the authors also discuss an iterative self-training procedure on top of PET which reports an impressive accuracy of $70.7\%$ on Yahoo Answers, which nearly approaches the performance of state-of-the-art supervised classification methods.

Topics

NLP (0.29)
Backend (0.27)
Machine_Learning (0.21)

Similar Articles

Creating the Whole Machine Learning Pipeline with PyCaret

By Medium - 2020-12-03

This tutorial covers the entire ML process, from data ingestion, pre-processing, model training, hyper-parameter fitting, predicting and storing the model for later use. We will complete all these…

30 Most Asked Machine Learning Questions Answered

By Medium - 2021-03-18

Machine Learning is the path to a better and advanced future. A Machine Learning Developer is the most demanding job in 2021 and it is going to increase by 20–30% in the upcoming 3–5 years. Machine…

How to Build a Machine Learning Model

By Medium - 2020-07-25

A Visual Guide to Learning Data Science

Fine-Tune Wav2Vec2 for English ASR in Hugging Face with 🤗 Transformers

By huggingface - 2021-03-12

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Learning Data Science From the Perspective of a Proficient Developer

By Medium - 2020-12-08

As you know, data science, and more specifically machine learning, is very much en vogue now, so guess what? I decided to enroll in a MOOC to become fluent in data science. But when you start with a…

How to put machine learning models into production

By Stack Overflow Blog - 2020-10-12

The goal of building a machine learning model is to solve a problem, and a machine learning model can only do so when it is in production and actively in use by consumers. As such, model deployment is ...