Learning to Generate Artificial Fovea Trajectories for Target Detection

By J. Urgen Schmidhuber, Rudolf Huber. 1991

Subfields: Computer Vision and Pattern Recognition, Reinforcement Learning

How researchers see this paper

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

In Computer Vision and Pattern Recognition

In Reinforcement Learning

Citations in the corpus

Learning from Dialogue after Deployment: Feed Yourself, Chatbot! (2019, background): “The general concepts of lifelong learning (Silver et al., 2013) and never-ending (language) learning (Carlson et al., 2010) are related to the topics discussed in this work, as is active learning (Tong and Koller, 2001) and predictive modeling (Schmidhuber and Huber, 1991).”
Dialog-based Language Learning (2016, methodology): “Finally, forward prediction models, which we make use of in this work, have been used for learning eye tracking [21], controlling robot arms [12] and vehicles [26], and action-conditional video prediction in atari games [16, 23].”
Spatial Transformer Networks (2015, background): “In this section we discuss the prior work related to the paper, covering the central ideas of modelling transformations with neural networks [15, 16, 36], learning and analysing transformation-invariant representations [4, 6, 10, 20, 22, 33], as well as attention and detection mechanisms for feature”
Recurrent World Models Facilitate Policy Evolution (2018, methodology): “Early work on RL for active vision trained an FNN to take the current image frame of a video sequence to predict the next frame [85], and use this predictive model to train a fovea-shifting control network trying to find targets in a visual scene.”
Action-Conditional Video Prediction using Deep Networks in Atari Games (2015, background): “The idea of building a predictive model for vision-based RL problems was introduced by Schmidhuber and Huber [27].”
Deep learning in neural networks: An overview (2014, methodology): “An RL active vision system used it to learn sequential shifts (saccades) of a fovea, to detect targets in visual scenes (Schmidhuber and Huber, 1991), thus learning to control selective attention.”
On the Binding Problem in Artificial Neural Networks (2020, background): “in the shape of a bounding-box (Stanley and Miikkulainen, 2004) or a fovea (Schmidhuber and Huber, 1991).”
World Models (2018, methodology): “Early work on RL for active vision trained an FNN to take the current image frame of a video sequence to predict the next frame (Schmidhuber & Huber, 1991), and use this predictive model to train a fovea-shifting control network trying to ﬁnd targets in a visual scene.”
Tagger: Deep Unsupervised Perceptual Grouping (2016, background): “Since recurrent neural networks (RNNs) are general purpose computers, they can in principle implement arbitrary computable types of temporary variable binding [23, 24] and unsupervised segmentation [22] and internal [24] and external attention [26].”
Model-Ensemble Trust-Region Policy Optimization (2018, neutral): “Although many previous studies have shown promising results on relatively simple domains (Nguyen and Widrow, 1990; Schmidhuber and Huber, 1991; Jordan and Rumelhart, 1992; Gal et al., 2016), so far their applications on more challenging domains have either required a combination with model-free tech”
Deep Networks with Internal Selective Attention through Feedback Connections (2014, background): “In the context of computer vision, RL has been shown to be able to learn saccades in visual scenes to learn selective attention [40, 41], learn feedback to lower levels [42, 43], and improve face recognition [44, 45].”
On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models (2015, methodology): “An RL active vision system, from 1991 [249], used this basic principle to learn sequential shifts (saccades) of a fovea to detect targets in a visual scene, thus learning a rudimentary version of selective attention.”
A Brief Survey of Deep Reinforcement Learning (2017, neutral): “manner, such as deciding where to look in an image to track [102], classify [70] or caption objects [141].”
Attentive Recurrent Comparators (2017, methodology): “Attention and (Recurrent) Neural Networks were combined in Schmidhuber & Huber (1991) to learn fovea trajectories.”
Attention for Fine-Grained Categorization (2014, background): “…beneﬁts of various attention mechanisms, though most of the work focuses on toy or more constrained environments, such as detecting simple shapes (Schmidhuber & Huber, 1991), tasks based on MNIST digits (Larochelle & Hinton, 2010; Bazzani et al., 2011; Denil et al., 2012; Ranzato, 2014; Mnih et…”
One Big Net For Everything (2018, methodology): “User-deﬁned goals were provided to the system by special “goal input vectors” that remained constant [79, Sec.”
Learning to Adapt in Dynamic, Real-World Environments through Meta-Reinforcement Learning (2018, background): “Our problem setting relates to meta-learning, a long-standing problem of interest in machine learning that is concerned with enabling artiﬁcial agents to efﬁciently learn new tasks by learning to learn (Thrun and Pratt, 1998; Schmidhuber and Huber, 1991; Naik and Mammone, 1992; Lake et al., 2015).”
Training Agents using Upside-Down Reinforcement Learning (2019, background): “There is also substantial prior work on learning reward or goal-conditional policies that directly produce actions [5, 6, 15, 35], but these rely either on a pre-trained world model, a dataset of goal and optimal policy parameters, or policy gradients for learning.”
Hindsight policy gradients (2017, background): “If the agent is required to achieve a specific goal during an episode, then such goal may be encoded as part of the states [2], possibly allowing the generalization of desired behavior to goals that were never encountered before.”
Learning to Adapt: Meta-Learning for Model-Based Control (2018, background): “Meta-Learning: The problem setting that we have been discussing also relates to meta-learning, a long-standing problem of interest in machine learning that is concerned with enabling artificial agents to efficiently learn new tasks by learning to learn [50, 45, 37, 23].”