Dynamic Neural Turing Machine with Continuous and Discrete Addressing Schemes

By Caglar Gulcehre, A. P. Sarath Chandar, Kyunghyun Cho, Yoshua Bengio. 2018

53 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Electrical and Electronic Engineering, Model Architectures

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. Kyunghyun Cho
2. Jiatao Gu
3. Graham Neubig
4. Mohit Bansal
5. Victor O. K. Li
6. Jianfeng Gao
7. Guodong Long
8. Yoshua Bengio

In Electrical and Electronic Engineering

1. Joelle Pineau
2. Prasanna Parthasarathi
3. Victor O. K. Li
4. Fei Wu
5. Tomas Lozano-Perez
6. Changsheng Xu
7. Hao Tang
8. Xiaoyuan Jing

In Model Architectures

1. Jiatao Gu
2. Victor O. K. Li
3. Guodong Long
4. Tianyi Zhou
5. Chengqi Zhang
6. Jing Jiang
7. Yi Yang
8. Yong Wang

Citations in the corpus

• Meta-Learning for Low-Resource Neural Machine Translation (2018, methodology): “Universal Lexical Representation (ULR) We tackle this issue by dynamically building a vocabulary specific to each language using a keyvalue memory network (Miller et al., 2016; Gulcehre et al., 2018), as was done successfully for low-resource machine translation recently by Gu et al.”
• DM-GAN: Dynamic Memory Generative Adversarial Networks for Text-To-Image Synthesis (2019, background): “Recently, memory network [5, 27] provides a new architecture to reason answers from memories more effectively using explicit storage and a notion of attention.”
• Extending Neural Generative Conversational Model using External Knowledge Sources (2018, methodology): “Although these models can represent the input context, the need for a dedicated external memory to remember information in context was pointed out and mechanisms were introduced in models like Memory Networks (Sukhbaatar et al., 2015; Bordes et al., 2016; Gulcehre et al., 2018), and the Neural Turin”
• Search Engine Guided Non-Parametric Neural Machine Translation (2017, background): “In a high level, the proposed TM-NMT first stores each target symbol of each retrieved translation pair into a key-value memory (Gulcehre et al., 2016b; Miller et al., 2016).”
• DF-GAN: Deep Fusion Generative Adversarial Networks for Text-to-Image Synthesis (2020, methodology): “Finally, DM-GAN [14] introduces the Memory Network [51], [52] to cope with the first aforementioned problem of stacked architecture.”
• Adversarial Text-to-Image Synthesis: A Review (2021, background): “DM-GAN [75] is an architecture based on dynamic memory networks [76, 77, 78, 79].”
• Graph Element Networks: adaptive, structured computation and memory (2019, background): “…systems that support read and write operations via attention (Ma & Pr´ıncipe, 2018; Graves et al., 2014; Kurach et al., 2015; Graves et al., 2016; G¨ulcehre et al., 2018) and especially to neural SLAM and neural map (Zhang et al., 2017; Parisotto & Salakhutdinov, 2017), which are also…”
• Reinforced Self-Attention Network: a Hybrid of Hard and Soft Attention for Sequence Modeling (2018, background): “Gulcehre et al. [ 2016 ] propose a dynamic neural Turing machine, which is implemented with both continuous (differentiable) and discrete (non-differentiable) memory read/write approaches.”
• DF-GAN: A Simple and Effective Baseline for Text-to-Image Synthesis (2020, methodology): “DM-GAN [60] introduces the Memory Network [10, 49] to refine fuzzy image contents when the initial images are not well generated in stacked architecture.”