Oja, Erkki

name

• Oja, Erkki

Authored works

• A nonlinear symmetrical PCA network learns true principal components
• Adaptation of a novelty filter system to a pattern subspace
• An analysis of convergence for a learning version of the subspace method
• Application of learning algorithms to selective learning and unlearning
• Asymptotic solutions of a class of matrix differential equations arising in neural network modelling
• Hahmontunnistustutkimusta Japanissa : raportti valtion teknistieteelliselle toimikunnalle opintomatkasta Japaniin 1983-1984
• Nonlinear PCA : algorithms and applications
• On stochastic approximation of the eigenvectors and eigenvalues of the expectation of a random matrix
• On the construction of projectors using products of elementary matrices
• On the convergence of an associative learning algorithm in the presence of noise
• On the generation of impulses in a second-order sensory neuron
• On the iterative solution of a linear matrix equation
• On the stochastic gradient ascent algorithm for generating the Karhunen-Loéve subspace
• S-orthogonalization in real vector spaces by means of a class of nonlinear matrix differential equations
• Studies of the convergence properties of adaptive orthogonalizing filters
• Subspace methods of pattern recognition
• The nonlinear PCA learning rule and signal separation : mathematical analysis

Works contributed to

• A fast fixed-point algorithm for independent component analysis
• A fast method for optimal adaptive data compression
• A model of cortical associative memory
• A note on a simple self-organizing process
• Adaptation of a linear system to a finite set of patterns occurring in an arbitrarily varying order
• An efficient method of computing a minimum spanning tree for a large complete graph
• Automaattinen tietämyksen omaksuminen ja koneoppiminen : tietotekniikan seminaari, kevät 1989
• Biennial report 2012-2013
• Can we solve the continuous Karhunen-Loève eigenproblem from the discrete data?
• Comparisons of attributed graph matching algorithms for computer vision
• Extended self-organizing map for curve detection
• Fast adaptive formation of orthogonalizing filters and associative memory in recurrent networks of neuron-like elements
• Generalized Randomized Hough Transform (GRHT)
• Hahmontunnistusalan sanastoa
• Image compression by neural networks : a comparison study
• Independent component analysis
• Independent component analysis by general non-linear Hebbian-like rules
• Motion detection using randomized Hough transform
• Neljännesvuosisata Hatutusta : hahmontunnistustutkimus Suomessa 1977-2002
• Neural network systems, techniques and applications in pattern recognition
• New methods for stochastic approximation of truncated Karhunen-Loève expansions
• Optimal FIR restoration filters with real or integer coefficients
• Randomized Hough transform (RHT) : theoretical analysis and extensions
• Recent versions of the hough transform and the randomized hough transform : overview and comparisons
• Recursive estimation of eigenvectors of correlation matrices
• Simple neuron models for independent component analysis
• Some comments on the subspace methods of classification
• Storage and processing of information in distributed associative memory systems
• The CARELIA simulator : a development and specification environment for neural networks

Works about Oja, Erkki

• Advances in independent component analysis and learning machines
• Kutsu kuulemaan niitä julkisia esitelmiä, jotka Lappeenrannan teknilliseen korkeakouluun nimitetyt tietojenkäsittelyopin professori Erkki Oja ja mekaanisen teknologian professori Tapio Lakso pitävät virkaanastujaisissaan marraskuun 29. päivänä 1988