Recursive Distributed Representations

By Jordan B. Pollack. 1990

Subfields: Representation & Self-Supervised Learning

How researchers see this paper

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

In Representation & Self-Supervised Learning

Citations in the corpus

Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank (2013, neutral): “, 2011b) and recursive auto-associate memories (Pollack, 1990).”
A Convolutional Neural Network for Modelling Sentences (2014, methodology): “A model that adopts a more general structure provided by an external parse tree is the Recursive Neural Network (RecNN) (Pollack, 1990; Küchler and Goller, 1996; Socher et al., 2011; Hermann and Blunsom, 2013).”
Neural Turing Machines (2014, background): “Considering this property, we do not feel that it is urgent or even necessarily valuable to build explicit parse trees to merge composite structures greedily (Pollack, 1990) (Socher et al., 2012) (Frasconi et al., 1998).”
Grounded Compositional Semantics for Finding and Describing Images with Sentences (2014, methodology): “RNN models (Pollack, 1990; Goller and Küchler, 1996; Socher et al., 2011b; Socher et al., 2011a) provided a novel way of combining word vectors for longer phrases that moved beyond simple averaging.”
Representation Learning: A Review and New Perspectives (2012, background): “on large minibatches (of several thousand examples) (Martens, 2010; Le et al., 2011a), with a variant of stochastic gradient descent recently winning an optimization challenge 14. Recursive networks (Pollack, 1990; Frasconi et al., 1997, 1998; Bottou, 2011; Socher et al., 2011) generalize recurrent ”
Deep learning in neural networks: An overview (2014, methodology): “Compare an early survey (Hinton, 1989) and the somewhat related Recursive Auto-Associative Memory (RAAM) (Melnik, Levy, & Pollack, 2000; Pollack, 1988, 1990), originally used to encode sequential linguistic structures of arbitrary size through a fixed number of hidden units.”
On the Binding Problem in Artificial Neural Networks (2020, background): “…inductive biases in neural networks that enable them to efficiently learn about symbols and the processes for manipulating them (examples of such an approach are abound, even in early connectionist research, e.g. Smolensky (1990); Pollack (1990); McMillan et al. (1992); Das and Mozer (1993)).”
Relational inductive biases, deep learning, and graph networks (2018, background): “…were developed in domains such as analogy-making, linguistic analysis, symbol manipulation, and other forms of relational reasoning (Smolensky, 1990; Hinton, 1990; Pollack, 1990; Elman, 1991; Plate, 1995; Eliasmith, 2013), as well as more integrative theories for how the mind works (Marcus, 2001).”
Deep Learning: Methods and Applications (2014, background): “, they can potentially perform powerful reasoning over structures, as argued in various historical publications as well as recent essays [38, 156, 286, 288, 292, 336, 335].”
Practical Recommendations for Gradient-Based Training of Deep Architectures (2012, methodology): “This idea has been generalized from sequences and 2-D images to arbitrary graphs with recursive neural networks or recursive graphical models (Pollack, 1990; Frasconi et al., 1998; Bottou, 2011; Socher et al., 2011), Markov Logic Networks (Richardson and Domingos, 2006) and relational learning (Geto”
From machine learning to machine reasoning (2011, background): “Such Recursive Auto-Associative Memory (RAAM) were proposed as a connectionist representation of infinite recursive structures (Pollack, 1990).”
Neural Tree Indexers for Text Understanding (2016, methodology): “A single layer network with tanh non-linearity was adopted in recursive auto-associate memories (Pollack, 1990) and recursive autoencoders (Socher et al.”
A Primer on Neural Network Models for Natural Language Processing (2015, methodology): “The recursive neural network (RecNN) abstraction (Pollack, 1990), popularized in NLP by Richard Socher and colleagues (Socher, Manning, & Ng, 2010; Socher, Lin, Ng, & Manning, 2011; Socher et al.”
Mapping Part-Whole Hierarchies into Connectionist Networks (1990, background): “If the hierarchical structure is highly overlearned, it is possible to train a network to recover the full description [23], but if the hierarchy is a novel one, the only way to move back up it without additional external guidance is to use some kind of content-addressable working memory for earlier”
RNNs Implicitly Implement Tensor Product Representations (2018, unknown): “These experiments suggest that RNNs only learn compositional representations when the task requires them.”
Neural-Symbolic Learning and Reasoning: A Survey and Interpretation (2017, background): “Additionally, only few systems used distributed representations—and even those which do typically suffer from information loss, and little processing abilities (Plate, 1995; Pollack, 1990; Stewart & Eliasmith, 2008).”
Convolutional-Recursive Deep Learning for 3D Object Classification (2012, methodology): “Lastly, recursive autoencoders have been introduced by Pollack [19] and Socher et al.”