Searcharxiv⌕ Search

arXiv subjects

Renata Avros

Publications and source records attributed to Renata Avros.

2 recordsLinked to original sources

Neural Classifiers as Error-Correcting Codes

A unified framework is proposed that characterizes supervised deep classification as the decoding of error-correcting codes in the Shannon sense. Cross-entropy optimization is shown to align softmax classifiers with maximum-likelihood decoding, where logits map to Hamming distances to class-specific codewords. This correspondence explains the emergence of standard divergence objectives (e.g., Kullback-Leibler, Jeffreys, Bregman) and provides a geometric basis for the stability of decision boundaries. Furthermore, the geometries of Neural Collapse and LeJEPA are identified as optimal constellations for decoding under isotropic noise. Unlike classical Error-Correcting Output Codes, deep networks are shown to learn effective codebooks in the feature space implicitly. This insight is leveraged to derive novel criteria for lossless quantization, improved knowledge distillation, and architectural regularization based on code redundancy. Finally, this perspective is extended to self-supervised encoders, offering a cohesive explanation for the geometry of latent representations in modern neural models.

cs.LG↗

Analysis of Metagenome Composition by the Method of Random Primers

Metagenome, a mixture of different genomes (as a rule, bacterial), represents a pattern, and the analysis of its composition is, currently, one of the challenging problems of bioinformatics. In the present study, the possibility of evaluating metagenome composition by DNA-marker methods is investigated. These methods are based on using primers, short nucleic acid fragments. Each primer picks out of the tested genome the fragment set specific just for this genome, which is called its spectrum (for the given primer) and is used for identifying the genome. The DNA-marker method, applied to a metagenome, also gives its spectrum, which, obviously, represents the union of the spectra of all genomes belonging to the metagenome. Thus each primer provides a projection of the genomes and of the metagenome onto the corresponding spectra set. Here we propose to apply the random projection (random primer) approach for analyzing metagenome composition and present some estimates of the method effectiveness for the case of Random Amplified Polymorphic DNA (RAPD) technology.

q-bio.QM↗