Hierarchical Neural Networks for Image Interpretation

  title={Hierarchical Neural Networks for Image Interpretation},
  author={Sven Behnke},
  booktitle={Lecture Notes in Computer Science},
  • Sven Behnke
  • Published in
    Lecture Notes in Computer…
    21 August 2003
  • Computer Science
I. Theory.- Neurobiological Background.- Related Work.- Neural Abstraction Pyramid Architecture.- Unsupervised Learning.- Supervised Learning.- II. Applications.- Recognition of Meter Values.- Binarization of Matrix Codes.- Learning Iterative Image Reconstruction.- Face Localization.- Summary and Conclusions. 
Layer-wise Learning of Feature Hierarchies
The reader is introduced to the basic concepts of deep learning, selected methods for incrementally learning a hierarchy of features from unlabeled inputs are discussed, and application examples from computer vision and speech recognition are presented.
Learning Iterative Binarization using Hierarchical Recurrent Networks
The binarization of matrix codes is investigated as an application of supervised learning of image processing tasks using a recurrent version of the Neural Pyramid to recognition of degraded matrix codes for which adaptive thresholding fails.
Face Localization in the Neural Abstraction Pyramid
This work proposes to use hierarchical neural networks with local recurrent connectivity to solve the localization of a face in an image, even in presence of complex backgrounds, difficult lighting, and noise.
Stacked Convolutional Auto-Encoders for Hierarchical Feature Extraction
A novel convolutional auto-encoder (CAE) for unsupervised feature learning that initializing a CNN with filters of a trained CAE stack yields superior performance on a digit and an object recognition benchmark.
On Fast Deep Nets for AGI Vision
This paper presents biologically inspired adaptive vision models, which greatly profit from recent advances in computing hardware, complementing recent progress in the AGI theory of mathematically optimal universal problem solvers.
Multi-column deep neural networks for image classification
On the very competitive MNIST handwriting benchmark, this method is the first to achieve near-human performance and improves the state-of-the-art on a plethora of common image classification benchmarks.
Deep Machine Learning with Spatio-Temporal Inference
This dissertation presents a novel DML architecture which is biologically-inspired in that all processing is performed hierarchically; all processing units are identical; and processing captures both spatial and temporal dependencies in the observations to organize and extract features suitable for supervised learning.
Fast image scanning with deep max-pooling convolutional neural networks
This work shows how dynamic programming can speedup the process by orders of magnitude, even when max-pooling layers are present.
High-Performance Neural Networks for Visual Object Classification
We present a fast, fully parameterizable GPU implementation of Convolutional Neural Network variants. Our feature extractors are neither carefully designed nor pre-wired, but rather learned in a
SSL Using Siamese Networks and Binary Classifier
This thesis presents several approaches for training a convolutional neural network using only unlabeled data, and combines these with the network called RotNet, which makes use of image rotations and therefore forces the network to learn rotation dependent features from the dataset.


Neural abstraction pyramid: a hierarchical image understanding architecture
  • Sven Behnke, R. Rojas
  • Computer Science
    1998 IEEE International Joint Conference on Neural Networks Proceedings. IEEE World Congress on Computational Intelligence (Cat. No.98CH36227)
  • 1998
A hierarchical neural architecture for image interpretation is proposed, which is based on image pyramids and cellular neural networks inspired by the principles of information processing found in the visual cortex, and first application, the binarization of handwriting, has been implemented and shown to improve the acceptance rate of an automatic ZIP-code recognition system without decreasing its reliability.
Natural image statistics and neural representation.
It has long been assumed that sensory neurons are adapted to the statistical properties of the signals to which they are exposed, but recent developments in statistical modeling have enabled researchers to study more sophisticated statistical models for visual images, to validate these models empirically against large sets of data, and to begin experimentally testing the efficient coding hypothesis.
A Simple and Fast Neural Network Approach to Stereovision
A neural network approach to stereovision is presented based on aliasing effects of simple disparity estimators and a fast coherence-detection scheme, which is fully parallel and non-iterative.
Hierarchical models of object recognition in cortex
A new hierarchical model consistent with physiological data from inferotemporal cortex that accounts for this complex visual task and makes testable predictions is described.
Probabilistic Visual Learning for Object Representation
An unsupervised technique for visual learning is presented, which is based on density estimation in high-dimensional spaces using an eigenspace decomposition and is applied to the probabilistic visual modeling, detection, recognition, and coding of human faces and nonrigid objects.
The CNN paradigm
A concise tutorial description of the cellular neural network (CNN) paradigm is given, along with a precise taxonomy, and it is shown how simply a wave-type partial differential equation can be generated.
Simplified neuron model as a principal component analyzer
  • E. Oja
  • Biology
    Journal of mathematical biology
  • 1982
A simple linear neuron model with constrained Hebbian-type synaptic modification is analyzed and a new class of unconstrained learning rules is derived. It is shown that the model neuron tends to
Neural Network-Based Face Detection
A neural network-based face detection system that arbitrates between multiple networks to improve performance over a single network using a bootstrap algorithm, which eliminates the difficult task of manually selecting non-face training examples.
Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects.
Results suggest that rather than being exclusively feedforward phenomena, nonclassical surround effects in the visual cortex may also result from cortico-cortical feedback as a consequence of the visual system using an efficient hierarchical strategy for encoding natural images.