no code implementations • 1 Jan 2021 • Sayan Ghosh, Eugene Laksana, Louis-Philippe Morency, Stefan Scherer
In this paper we propose the IMA (Importance-based Multimodal Autoencoder) model, a scalable model that learns modality importances and robust multimodal representations through a novel cross-covariance based loss function.
no code implementations • 28 Jul 2020 • David Ledbetter, Eugene Laksana, Melissa Aczon, Randall Wetzel
This work presents input data perseveration as a method of training and deploying an RNN model to make its predictions more responsive to newly acquired information: input data is replicated during training and deployment.
no code implementations • 1 Apr 2019 • Eugene Laksana, Melissa Aczon, Long Ho, Cameron Carlin, David Ledbetter, Randall Wetzel
Electronic Medical Records (EMR) are a rich source of patient information, including measurements reflecting physiologic signs and administered therapies.
no code implementations • ACL 2017 • Sayan Ghosh, Mathieu Chollet, Eugene Laksana, Louis-Philippe Morency, Stefan Scherer
Human verbal communication includes affective messages which are conveyed through use of emotionally colored words.
no code implementations • 15 Nov 2015 • Sayan Ghosh, Eugene Laksana, Louis-Philippe Morency, Stefan Scherer
Experiments on a well-established real-life speech dataset (IEMOCAP) show that the learnt representations are comparable to state of the art feature extractors (such as voice quality features and MFCCs) and are competitive with state-of-the-art approaches at emotion and dimensional affect recognition.