Nov 2, 2020 · Our aim is to obtain a classifier that performs as well as possible on examples of each "common class" that occurs with frequency above a given ...

Exemplar Guided Active Learning (EGAL) is a method we developed to economically annotate training data for tasks with extremely skewed label distributions, ...

Once we have found examples of common classes, we switch to more standard active learning methods to find additional examples to reduce classifier uncertainty.

Summary and Contributions: The paper proposes a framework that identifies rare and common classes from the dataset using word embedding techniques and ...

We present the Exemplar Guided Active Learning algorithm that leverages the embedding spaces of large scale language models to drastically improve active ...

Sep 9, 2024 · Our aim is to obtain a classifier that performs as well as possible on examples of each "common class" that occurs with frequency above a given ...

Dec 6, 2020 · We describe an active learning approach that (1) explicitly searches for rare classes by leveraging the contextual embedding spaces provided by ...

Papertalk is an open-source platform where scientists share video presentations about their newest scientific results - and watch, like + discuss them.

Jason S. Hartford, Kevin Leyton-Brown, Hadas Raviv, Dan Padnos, Shahar Lev, Barak Lenz: Exemplar Guided Active Learning. CoRR abs/2011.01285 (2020).

Dec 6, 2020 · Neural Information Processing Systems (NeurIPS) is a multi-track machine learning and computational neuroscience conference that includes ...