Recursive Hetero-associative Memories for Translation

By Mikel L. Forcada, Ramón P. Ñeco. 1997

How researchers see this paper

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

In Natural Language Processing

Citations in the corpus

Deep Recurrent Models with Fast-Forward Connections for Neural Machine Translation (2016, background): “In this task, the likelihood p(y | x,θ) of the target sequence will be maximized (Forcada and Ñeco, 1997) with parameter θ to learn:”
Neural Machine Translation by Jointly Learning to Align and Translate (2014, background): “Recently, a number of papers have proposed the use of neural networks to directly learn this conditional distribution (see, e.g., Kalchbrenner and Blunsom, 2013; Cho et al. , 2014a; Sutskever et al. , 2014; Cho et al. , 2014b; Forcada and ˜ Neco, 1997).”
A Primer on Neural Network Models for Natural Language Processing (2015, neutral): “An example of using the ranking log loss can be found in the work of Gao et al. (2014). A variation of the ranking log loss allowing for a different margin for the negative and positive class is given in work by dos Santos et al.”
The University of Cambridge’s Machine Translation Systems for WMT18 (2018, background): “Encoder-decoder networks (Pollack, 1990; Chris-man, 1991; Forcada and ˜Neco, 1997; Kalchbren-ner and Blunsom, 2013) are the current prevailing architecture for neural machine translation (NMT).”
Non-Monotonic Sequential Text Generation (2019, background): “…proposed binary tree generating policy, as it has been shown to encode a variable-sized input and predict a structured output effectively (Cleeremans et al., 1989; Forcada & ˜Neco, 1997; Sutskever et al., 2014; Cho et al., 2014b; Tai et al., 2015; Bronstein et al., 2017; Battaglia et al., 2018).”
On internal language representations in deep learning: an analysis of machine translation and speech recognition (2018, methodology): “The encoder maps a source sentence to a vector representation, which the decoder (19)As noted by Koehn [182], some of these models are remarkably similar to the modem encoder-decoder approach to neural machine translation; see for example figure 1 in [109].”
Asynchronous translations with recurrent neural nets (1997, methodology): “More recently [8] we have devised a related approach which we call recursive heteroassociative memories (RHAM), which is computationally equivalent to the RAAM translators of [5] and [6] but simpler because it does not require learning a complete autoassociator for the strings in both languages.”