An official website of the United States government
Summary Leaf vein network geometry can predict levels of resource transport, defence and mechanical support that operate at different spatial scales. However, it is challenging to quantify network architecture across scales due to the difficulties both in segmenting networks from images and in extracting multiscale statistics from subsequent network graph representations.Here we developed deep learning algorithms using convolutional neural networks (CNNs) to automatically segment leaf vein networks. Thirty‐eight CNNs were trained on subsets of manually defined ground‐truth regions from >700 leaves representing 50 southeast Asian plant families. Ensembles of six independently trained CNNs were used to segment networks from larger leaf regions (c. 100 mm2). Segmented networks were analysed using hierarchical loop decomposition to extract a range of statistics describing scale transitions in vein and areole geometry.The CNN approach gave a precision‐recall harmonic mean of 94.5% ± 6%, outperforming other current network extraction methods, and accurately described the widths, angles and connectivity of veins. Multiscale statistics then enabled the identification of previously undescribed variation in network architecture across species.We provide aLeafVeinCNNsoftware package to enable multiscale quantification of leaf vein networks, facilitating the comparison across species and the exploration of the functional significance of different leaf vein architectures.
more »
« less
This content will become publicly available on August 29, 2026
Summary statistics of learning link changing neural representations to behavior
How can we make sense of large-scale recordings of neural activity across learning? Theories of neural network learning with their origins in statistical physics offer a potential answer: for a given task, there are often a small set of summary statistics that are sufficient to predict performance as the network learns. Here, we review recent advances in how summary statistics can be used to build theoretical understanding of neural network learning. We then argue for how this perspective can inform the analysis of neural data, enabling better understanding of learning in biological and artificial neural networks.
more »
« less
- PAR ID:
- 10650392
- Publisher / Repository:
- Frontiers Media S.A., Switzerland
- Date Published:
- Journal Name:
- Frontiers in Neural Circuits
- Volume:
- 19
- ISSN:
- 1662-5110
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract We present a proof-of-concept simulation-based inference on Ωmandσ8from the Sloan Digital Sky Survey (SDSS) Baryon Oscillation Spectroscopic Survey (BOSS) LOWZ Northern Galactic Cap (NGC) catalog using neural networks and domain generalization techniques without the need of summary statistics. Using rapid light-cone simulations L-picola, mock galaxy catalogs are produced that fully incorporate the observational effects. The collection of galaxies is fed as input to a point cloud-based network,Minkowski-PointNet. We also add relatively more accurate Gadgetmocks to obtain robust and generalizable neural networks. By explicitly learning the representations that reduce the discrepancies between the two different data sets via the semantic alignment loss term, we show that the latent space configuration aligns into a single plane in which the two cosmological parameters form clear axes. Consequently, during inference, the SDSS BOSS LOWZ NGC catalog maps onto the plane, demonstrating effective generalization and improving prediction accuracy compared to non-generalized models. Results from the ensemble of 25 independently trained machines find Ωm= 0.339 ± 0.056 andσ8= 0.801 ± 0.061, inferred only from the distribution of galaxies in the light-cone slices without relying on any indirect summary statistics. A single machine that best adapts to the Gadgetmocks yields a tighter prediction of Ωm= 0.282 ± 0.014 andσ8= 0.786 ± 0.036. We emphasize that adaptation across multiple domains can enhance the robustness of the neural networks in observational data.more » « less
-
SNP heritability, the proportion of phenotypic variation explained by genotyped SNPs, is an important parameter in understanding the genetic architecture underlying various diseases and traits. Methods that aim to estimate SNP heritability from individual genotype and phenotype data are limited by their ability to scale to Biobank-scale data sets and by the restrictions in access to individual-level data. These limitations have motivated the development of methods that only require summary statistics. Although the availability of publicly accessible summary statistics makes them widely applicable, these methods lack the accuracy of methods that utilize individual genotypes. Here we present a SUMmary-statistics-based Randomized Haseman-Elston regression (SUM-RHE), a method that can estimate the SNP heritability of complex phenotypes with accuracies comparable to approaches that require individual genotypes, while exclusively relying on summary statistics. SUM-RHE employs Genome-Wide Association Study (GWAS) summary statistics and statistics obtained on a reference population, which can be efficiently estimated and readily shared for public use. Our results demonstrate that SUM-RHE obtains estimates of SNP heritability that are substantially more accurate compared with other summary statistic methods and on par with methods that rely on individual-level data.more » « less
-
null (Ed.)Abstract Recurrent neural networks have led to breakthroughs in natural language processing and speech recognition. Here we show that recurrent networks, specifically long short-term memory networks can also capture the temporal evolution of chemical/biophysical trajectories. Our character-level language model learns a probabilistic model of 1-dimensional stochastic trajectories generated from higher-dimensional dynamics. The model captures Boltzmann statistics and also reproduces kinetics across a spectrum of timescales. We demonstrate how training the long short-term memory network is equivalent to learning a path entropy, and that its embedding layer, instead of representing contextual meaning of characters, here exhibits a nontrivial connectivity between different metastable states in the underlying physical system. We demonstrate our model’s reliability through different benchmark systems and a force spectroscopy trajectory for multi-state riboswitch. We anticipate that our work represents a stepping stone in the understanding and use of recurrent neural networks for understanding the dynamics of complex stochastic molecular systems.more » « less
-
null (Ed.)ABSTRACT Emission from the interstellar medium can be a significant contaminant of measurements of the intensity and polarization of the cosmic microwave background (CMB). For planning CMB observations, and for optimizing foreground-cleaning algorithms, a description of the statistical properties of such emission can be helpful. Here, we examine a machine learning approach to inferring the statistical properties of dust from observational data. In particular, we apply a type of neural network called a variational autoencoder (VAE) to maps of the intensity of emission from interstellar dust as inferred from Planck sky maps and demonstrate its ability to (i) simulate new samples with similar summary statistics as the training set, (ii) provide fits to emission maps withheld from the training set, and (iii) produce constrained realizations. We find VAEs are easier to train than another popular architecture: that of generative adversarial networks, and are better suited for use in Bayesian inference.more » « less
