Rapid adaptation for deep neural networks through multi-task learning

By Zhen Huang, Jinyu Li, Sabato Marco Siniscalchi, I-Fan Chen, Ji Wu, Chin-Hui Lee. 2015

69 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Jinyu Li
2. Dong Yu
3. Li Deng
4. Yifan Gong
5. Daniel Povey
6. Shinji Watanabe
7. Sanjeev Khudanpur
8. Yoshua Bengio

In Signal Processing

1. Jinyu Li
2. Dong Yu
3. Yifan Gong
4. Li Deng
5. Shinji Watanabe
6. Chin-Hui Lee
7. Daniel Povey
8. Sanjeev Khudanpur

Citations in the corpus

• Salvaging Federated Learning by Local Adaptation (2020, background): “, 2015), but (Huang et al., 2015) makes the connection between personalization and multi-task learning.”
• Beyond Shared Hierarchies: Deep Multitask Learning through Soft Layer Ordering (2017, background): “…al., 2014), natural language (Collobert and Weston, 2008; Dong et al., 2015; Hashimoto et al., 2016; Liu et al., 2015a; Luong et al., 2016), speech (Huang et al., 2013; 2015; Seltzer and Droppo, 2013; Wu et al., 2015), and reinforcement learning (Devin et al., 2016; Fernando et al., 2017;…”
• Multi-Task Learning as Multi-Objective Optimization (2018, background): “…processing (Collobert and Weston, 2008; Dong et al., 2015; Liu et al., 2015a; Luong et al., 2015; Hashimoto et al., 2017), speech processing (Huang et al., 2013; Seltzer and Droppo, 2013; Huang et al., 2015), and even seemingly unrelated domains over multiple modalities (Kaiser et al., 2017).”
• Unsupervised adaptation with domain separation networks for robust speech recognition (2017, methodology): “To address this issue, regularization-based approaches are proposed in [9, 10, 11, 12] to regularize the neuron output distributions or the model parameters.”
• Recent progresses in deep learning based acoustic models (2017, methodology): “In [100], a multi-task learning (MTL) framework was proposed by adding auxiliary monophone classification as the second task in addition to the primary seone classification task.”
• Investigation of transfer learning for ASR using LF-MMI trained neural networks (2017, methodology): “LHNbased adaptation is compared with multi-task learning in [16].”
• Bone-conducted speech enhancement using deep denoising autoencoder (2018, background): “In the literature, the limited generalization capability is a common issue in many speech-related applications, such as ASR (Li et al., 2014; Huang et al., 2015), speaker recognition (Kolbœk et al.”
• Pareto Multi-Task Learning (2019, background): “Many MTL approaches have been proposed in the past, and they achieve great performances in many areas such as computer vision [2], natural language processing [3] and speech recognition [4].”
• Multi-objective learning and mask-based post-processing for deep neural network based speech enhancement (2017, methodology): “In fact, it was similar to the multi-task learning [26], but different from the multi-task learning in recent DNN-based speech recognition [27, 28] with only one input feature type.”
• Evolutionary architecture search for deep multitask networks (2018, background): “sthatareusedindifferentways at different depths for different tasks. Deep MTL has extended these ideas to domains where deep learning thrives, including vision [2, 17, 26, 31, 33, 35, 49, 51], speech [11,12,17,39,48],naturallanguageprocessing[4,6,9,17,24,27,50], and reinforcement learning [5, 15, 47”
• Conditional Teacher-student Learning (2019, methodology): “Many techniques have been proposed for speaker adaptation of deep acoustic models, such as regularization-based [25, 26, 27], transformation-based [28, 29], singular value decomposition-based”
• Hierarchical Bayesian combination of plug-in maximum a posteriori decoders in deep neural networks-based speech recognition and speaker adaptation (2017, methodology): “For adaptation of the DNN based ASR system in this paper, an affine transform network, namely the LHN, is inserted into the original neural architecture (Gemello et al., 2007) and MAP LHN adaption follows (Huang et al., 2015b).”
• Speaker-Invariant Training Via Adversarial Learning (2018, background): “Therefore, speaker adaptation meth-ods are widely used to boost the recognition system performance [3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13].”
• Factorized Hidden Layer Adaptation for Deep Neural Network Based Acoustic Modeling (2016, methodology): “Some techniques deal with the sparsity issue of the adaptation data by performing monophone adaptation with multitask learning [52] or using a hierarchy of output layers [53] or using structured output layers [54].”