The “independent components” of natural scenes are edge filters

1997/12/01 by Anthony J. Bell, Terrence J. Sejnowski, T. J. Sejnowski · 2,273 citations
Computer Science · Mathematics · Neuroscience · Psychology · #Artificial intelligence #Blind Source Separation Techniques #Blind signal separation #Computer science #Computer vision #Filter (signal processing) #Independent component analysis #Infomax #Mathematics #Maximization #Neural dynamics and brain function #Pattern recognition (psychology) #Psychology #Receptive field #Representation (politics) #Visual cortex #Visual perception and processing mechanisms

paper · pdf · doi:10.1016/s0042-6989(97)00121-1

published in Vision Research 37(23), 3327-3338 (Elsevier BV)

openalex publication_date 1997/12/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/01

Cited by

• Clustering Learning for Robotic Vision
• A Tutorial on Principal Component Analysis
• Towards Machine Intelligence
• Independent component analysis: algorithms and applications
• Neural tuning and representational geometry
• A survey on modern trainable activation functions
• Feedback inhibition shapes emergent computational properties of cortical microcircuit motifs
• Representation Learning: A Review and New Perspectives
• Learning to See by Moving
• Object Detection in 20 Years: A Survey
• Explicit Disentanglement of Appearance and Perspective in Generative Models
• Third-Order Edge Statistics: Contour Continuation, Curvature, and Cortical Connections
• Dissociating the pre-activation of word meaning and form during sentence comprehension: Evidence from EEG representational similarity analysis
• Distractor anticipation during working memory is associated with theta and beta oscillations across spatial scales
• Exact feature probabilities in images with occlusion
• Synergy from Silence in a Combinatorial Neural Code
• Representation Learning: A Review and New Perspectives
• Behavioral alignment as an organizing principle in sensory coding
• Non-negative matrix factorization with sparseness constraints
• On Feature Decorrelation in Self-Supervised Learning
• Natural statistics of binaural sounds
• Generative Image Modeling Using Spatial LSTMs
• EigenGame: PCA as a Nash Equilibrium
• On the achievability of blind source separation for high-dimensional nonlinear source mixtures
• Modeling Information Flow Through Deep Neural Networks
• Free Component Analysis: Theory, Algorithms & Applications
• Sparse and redundant signal representations for x-ray computed tomography
• Dynamic mode decomposition of GRACE satellite data
• Linear and non-linear Modified Gravity forecasts with future surveys
• Fast and robust fixed-point algorithms for independent component analysis
• Shared Representation Learning for Heterogeneous Face Recognition
• The effect of wording on message propagation: Topic- and author-controlled natural experiments on Twitter
• Learning Mid-Level Auditory Codes from Natural Sound Statistics
• A Generative Model of Natural Texture Surrogates
• Blind Source Separation: Fundamentals and Recent Advances (A Tutorial Overview Presented at SBrT-2001)
• Efficient Parametric Projection Pursuit Density Estimation
• What Is the Relation Between Slow Feature Analysis and Independent Component Analysis?
• The Generalized Spike Process, Sparsity, and Statistical Independence
• Deep Learning in Spiking Neural Networks
• How Convolutional Neural Network Architecture Biases Learned Opponency and Colour Tuning
• An improvement on LSB+ method
• Neuroscience-inspired online unsupervised learning algorithms
• Sparse Gaussian ICA
• Sparsity vs. Statistical Independence in Adaptive Signal Representations: A Case Study of the Spike Process
• Finite Versus Infinite Neural Networks: an Empirical Study
• Feature Whitening via Gradient Transformation for Improved Convergence
• The perceptual primacy of feeling: Affectless visual machines explain a majority of variance in human visually evoked affect
• Selective inference for k-means clustering
• A New Basis for Sparse Principal Component Analysis
• The local low-dimensionality of natural images
• Discriminately Decreasing Discriminability with Learned Image Filters
• Stochastic Whitening Batch Normalization
• Learning to Factor Policies and Action-Value Functions: Factored Action Space Representations for Deep Reinforcement learning
• Face recognition by elastic bunch graph matching
• Self-Similarity Properties of Natural Images Resemble Those of Turbulent Flows
• Optimal Whitening and Decorrelation
• How Behavioral Constraints May Determine Optimal Sensory Representations
• Feature detection using spikes: The greedy approach
• Temporal Autoencoding Restricted Boltzmann Machine
• Reconstruction of Natural Scenes from Ensemble Responses in the Lateral Geniculate Nucleus
• Quantum computing in neural networks
• Natural Images from the Birthplace of the Human Eye
• Human eye-head co-ordination in natural exploration
• Derivatives and inverse of cascaded linear+nonlinear neural models
• Hocus-Socus: An Error Catastrophe for Complex Hebbian Learning Implies\n Neocortical Proofreading
• Correlation-invariant synaptic plasticity
• Generating adaptive and robust filter sets using an unsupervised learning framework
• A critique of pure learning and what artificial neural networks can learn from animal brains
• Functionally independent components of the late positive event-related potential during visual spatial attention. [europepmc]
• Reconstruction of natural scenes from ensemble responses in the lateral geniculate nucleus. [europepmc]
• Natural stimulation of the nonclassical receptive field increases information transmission efficiency in V1. [europepmc]
• Isolation of relevant visual features from random stimuli for cortical complex cells. [europepmc]
• Synergy, redundancy, and independence in population codes. [europepmc]
• Face recognition by independent component analysis. [europepmc]
• SUN: A Bayesian framework for saliency using natural statistics. [europepmc]
• Muscle synergies as a predictive framework for the EMG patterns of new hand postures. [europepmc]
• Neural representation of natural images in visual area V2. [europepmc]
• Statistically optimal perception and learning: from behavior to neural representations. [europepmc]
• Imaging Brain Dynamics Using Independent Component Analysis. [europepmc]
• Spontaneous cortical activity reveals hallmarks of an optimal internal model of the environment. [europepmc]
• Defining the computational structure of the motion detector in Drosophila. [europepmc]
• Efficient coding of spatial information in the primate retina. [europepmc]
• Inhibitory interneurons decorrelate excitatory cells to drive sparse code formation in a spiking model of V1. [europepmc]
• Population code in mouse V1 facilitates readout of natural scenes through increased sparseness. [europepmc]
• Efficient sensory encoding and Bayesian inference with heterogeneous neural populations. [europepmc]
• Benefits of pathway splitting in sensory coding. [europepmc]
• Mapping of functionally characterized cell classes onto canonical circuit operations in primate prefrontal cortex. [europepmc]
• Flexible gating of contextual influences in natural vision. [europepmc]
• The relationship between spatial configuration and functional connectivity of brain regions. [europepmc]
• A critique of pure learning and what artificial neural networks can learn from animal brains. [europepmc]
• Morphology and gene expression profiling provide complementary information for mapping cell state. [europepmc]
• Epigenomic dissection of Alzheimer's disease pinpoints causal variants and reveals epigenome erosion. [europepmc]
• Single-cell multiregion dissection of Alzheimer's disease. [europepmc]