arXiv subjects

Michael Biehl

Publications and source records attributed to Michael Biehl.

21 records · Page 2Linked to original sources

arXiv2000

Singularity spectra of rough growing surfaces from wavelet analysis

We apply the wavelet transform modulus maxima (WTMM) method to the analysis of simulated MBE-grown surfaces. In contrast to the structure function approach commonly used in the literature, this new method permits an investigation of the complete singularity spectrum. We focus on a kinetic Monte-Carlo model with Arrhenius dynamics, which in particular takes into consideration the process of thermally activated desorption of particles. We find a wide spectrum of Hoelder exponents, which reflects the multiaffine surface morphology. Although our choice of parameters yields small desorption rates (< 3 %), we observe a dramatic change in the singularity spectrum, which is shifted towards smaller Hoelder exponents. Our results offer a mathematical foundation of anomalous scaling: We identify the global exponent alpha_g with the Hoelder exponent which maximizes the singularity soectrum.

cond-mat.stat-mech↗

arXiv1999

Noisy regression and classification with continuous multilayer networks

We investigate zero temperature Gibbs learning for two classes of unrealizable rules which play an important role in practical applications of multilayer neural networks with differentiable activation functions: classification problems and noisy regression problems. Considering one step of replica symmetry breaking, we surprisingly find that for sufficiently large training sets the stable state is replica symmetric even though the target rule is unrealizable. Further, the classification problem is shown to be formally equivalent to the noisy regression problem.

cond-mat.dis-nn↗

arXiv1998

Statistical physics and practical training of soft-committee machines

Equilibrium states of large layered neural networks with differentiable activation function and a single, linear output unit are investigated using the replica formalism. The quenched free energy of a student network with a very large number of hidden units learning a rule of perfectly matching complexity is calculated analytically. The system undergoes a first order phase transition from unspecialized to specialized student configurations at a critical size of the training set. Computer simulations of learning by stochastic gradient descent from a fixed training set demonstrate that the equilibrium results describe quantitatively the plateau states which occur in practical training procedures at sufficiently small but finite learning rates.

cond-mat.dis-nn↗