On handling negative transfer and imbalanced distributions in multiple source transfer learning
By Jing Gao, Liang Ge, Kang Li, Hung Quoc Ngo, Aidong Zhang. 2014
131 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
- Yulia Tsvetkov
- Graham Neubig
- Orhan Firat
- Zirui Wang
- Mohit Bansal
- Yuan Cao
- Steven C. H. Hoi
- Dongrui Wu
In Signal Processing
- Doyen Sahoo
- Steven C. H. Hoi
- Lingfei Deng
- Jaime G. Carbonell
- Peilin Zhao
- Dongrui Wu
- Yuan Cao
- Yulia Tsvetkov
Citations in the corpus
- Efficient Meta Lifelong-Learning with Limited Memory (2020, background): “In addition, as prior work have shown transferring from diversely related sources can hurt performance in the target (Ge et al., 2014; Wang and Carbonell, 2018), we study if transferring from Enc-Dec Replay MbPA++ Meta-MbPA”
- Domain Adaptation for Visual Applications: A Comprehensive Survey (2017, methodology): “W of each source cluster is computed by comparing the source and target clusters and is used by the final target learner to attenuate the effects of negative transfer. Similarly, Gao et al. propose in [96] a Locally Weighted Ensemble (LWE) framework to combine multiple source models. They approximate”
- Overcoming Negative Transfer: A Survey (2020, background): “And similar negative effects have also been observed in various settings including imbalanced distributions [33] and open set transfer learning [49].”
- Characterizing and Avoiding Negative Transfer (2018, background): “Some more recent work [7, 10, 3] has also observed similar negative impact while performing transfer learning on more complex tasks under different settings.”
- Online Learning: A Comprehensive Survey (2018, background): “Other Related Work Multi-source Online Transfer Learning (MSOTL) (Ge et al., 2013; Tommasi et al., 2012) solves a more challenging problem where k classifiers h1, ...hk are provided by k sources.”
- Gradient Vaccine: Investigating and Improving Multi-task Optimization in Massively Multilingual Models (2020, background): “As prior work have shown training with unbalanced tasks can be prone to negative interference (Ge et al., 2014; Wang & Carbonell, 2018), we study how to mitigate it in large models trained with highly unbalanced and massive-scale dataset.”