Study Demonstrates Predictive Coding Achieves Greater Sample Efficiency Than Backpropagation in Neural Networks
Researchers have derived analytical expressions explaining why Predictive Coding (PC) achieves greater sample efficiency than Backpropagation (BP) in neural network training. Using a metric called 'target alignment,' the study shows PC's advantage is most pronounced in deep, narrow, and pre-trained networks. The findings provide a mechanistic foundation that could guide how PC-based models are designed and parameterized for optimal learning.
A new preprint posted to arXiv introduces a theoretical framework for understanding the sample efficiency advantage of Predictive Coding (PC) over Backpropagation (BP), two competing accounts of how neural networks — and potentially biological cortical circuits — learn. The authors propose a metric called 'target alignment,' which quantifies how closely a network's output change aligns with its prediction error during training. Analytical expressions for target alignment were derived and empirically validated in Deep Linear Networks, with results showing PC consistently outperforms BP, especially in deep, narrow, and pre-trained architectures. The study also identifies exact conditions under which PC achieves guaranteed optimal target alignment. Experiments on full training trajectories of both linear and non-linear models suggest the predicted benefits hold even when some theoretical assumptions are relaxed. The work aims to move the field beyond empirical comparisons of PC and BP toward a principled, mechanistic understanding of their differences.
What's missing
As a preprint, this work has not yet undergone formal peer review. The study's analytical results are derived primarily for Deep Linear Networks, and while the authors report that benefits persist in non-linear models, the extent to which the theoretical guarantees generalize to large-scale, real-world architectures remains an open question.
What different sources said
- arXiv cs.LGCenter
Understanding Sample Efficiency in Predictive Coding
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.