Alert button
Picture for Herman Kamper

Herman Kamper

Alert button

Vector-quantized neural networks for acoustic unit discovery in the ZeroSpeech 2020 challenge

May 19, 2020
Benjamin van Niekerk, Leanne Nortje, Herman Kamper

Figure 1 for Vector-quantized neural networks for acoustic unit discovery in the ZeroSpeech 2020 challenge
Figure 2 for Vector-quantized neural networks for acoustic unit discovery in the ZeroSpeech 2020 challenge
Figure 3 for Vector-quantized neural networks for acoustic unit discovery in the ZeroSpeech 2020 challenge
Figure 4 for Vector-quantized neural networks for acoustic unit discovery in the ZeroSpeech 2020 challenge
Viaarxiv icon

Analyzing autoencoder-based acoustic word embeddings

Apr 03, 2020
Yevgen Matusevych, Herman Kamper, Sharon Goldwater

Figure 1 for Analyzing autoencoder-based acoustic word embeddings
Figure 2 for Analyzing autoencoder-based acoustic word embeddings
Figure 3 for Analyzing autoencoder-based acoustic word embeddings
Figure 4 for Analyzing autoencoder-based acoustic word embeddings
Viaarxiv icon

Unsupervised feature learning for speech using correspondence and Siamese networks

Mar 28, 2020
Petri-Johan Last, Herman A. Engelbrecht, Herman Kamper

Figure 1 for Unsupervised feature learning for speech using correspondence and Siamese networks
Figure 2 for Unsupervised feature learning for speech using correspondence and Siamese networks
Figure 3 for Unsupervised feature learning for speech using correspondence and Siamese networks
Figure 4 for Unsupervised feature learning for speech using correspondence and Siamese networks
Viaarxiv icon

Masakhane -- Machine Translation For Africa

Mar 13, 2020
Iroro Orife, Julia Kreutzer, Blessing Sibanda, Daniel Whitenack, Kathleen Siminyu, Laura Martinus, Jamiil Toure Ali, Jade Abbott, Vukosi Marivate, Salomon Kabongo, Musie Meressa, Espoir Murhabazi, Orevaoghene Ahia, Elan van Biljon, Arshath Ramkilowan, Adewale Akinfaderin, Alp Öktem, Wole Akin, Ghollah Kioko, Kevin Degila, Herman Kamper, Bonaventure Dossou, Chris Emezue, Kelechi Ogueji, Abdallah Bashir

Figure 1 for Masakhane -- Machine Translation For Africa
Viaarxiv icon

Multilingual acoustic word embedding models for processing zero-resource languages

Feb 21, 2020
Herman Kamper, Yevgen Matusevych, Sharon Goldwater

Figure 1 for Multilingual acoustic word embedding models for processing zero-resource languages
Figure 2 for Multilingual acoustic word embedding models for processing zero-resource languages
Figure 3 for Multilingual acoustic word embedding models for processing zero-resource languages
Figure 4 for Multilingual acoustic word embedding models for processing zero-resource languages
Viaarxiv icon

Deep motion estimation for parallel inter-frame prediction in video compression

Dec 11, 2019
André Nortje, Herman A. Engelbrecht, Herman Kamper

Figure 1 for Deep motion estimation for parallel inter-frame prediction in video compression
Figure 2 for Deep motion estimation for parallel inter-frame prediction in video compression
Figure 3 for Deep motion estimation for parallel inter-frame prediction in video compression
Figure 4 for Deep motion estimation for parallel inter-frame prediction in video compression
Viaarxiv icon

BINet: a binary inpainting network for deep patch-based image compression

Dec 11, 2019
André Nortje, Willie Brink, Herman A. Engelbrecht, Herman Kamper

Figure 1 for BINet: a binary inpainting network for deep patch-based image compression
Figure 2 for BINet: a binary inpainting network for deep patch-based image compression
Figure 3 for BINet: a binary inpainting network for deep patch-based image compression
Figure 4 for BINet: a binary inpainting network for deep patch-based image compression
Viaarxiv icon

If dropout limits trainable depth, does critical initialisation still matter? A large-scale statistical analysis on ReLU networks

Oct 13, 2019
Arnu Pretorius, Elan van Biljon, Benjamin van Niekerk, Ryan Eloff, Matthew Reynard, Steve James, Benjamin Rosman, Herman Kamper, Steve Kroon

Figure 1 for If dropout limits trainable depth, does critical initialisation still matter? A large-scale statistical analysis on ReLU networks
Figure 2 for If dropout limits trainable depth, does critical initialisation still matter? A large-scale statistical analysis on ReLU networks
Figure 3 for If dropout limits trainable depth, does critical initialisation still matter? A large-scale statistical analysis on ReLU networks
Figure 4 for If dropout limits trainable depth, does critical initialisation still matter? A large-scale statistical analysis on ReLU networks
Viaarxiv icon

On the expected behaviour of noise regularised deep neural networks as Gaussian processes

Oct 12, 2019
Arnu Pretorius, Herman Kamper, Steve Kroon

Figure 1 for On the expected behaviour of noise regularised deep neural networks as Gaussian processes
Figure 2 for On the expected behaviour of noise regularised deep neural networks as Gaussian processes
Figure 3 for On the expected behaviour of noise regularised deep neural networks as Gaussian processes
Figure 4 for On the expected behaviour of noise regularised deep neural networks as Gaussian processes
Viaarxiv icon