arXiv subjects

Chris Rohlfs

Publications and source records attributed to Chris Rohlfs.

4 recordsLinked to original sources

arXiv2024

Generalization in Neural Networks: A Broad Survey

This paper reviews concepts, modeling approaches, and recent findings along a spectrum of different levels of abstraction of neural network models including generalization across (1) Samples, (2) Distributions, (3) Domains, (4) Tasks, (5) Modalities, and (6) Scopes. Strategies for (1) sample generalization from training to test data are discussed, with suggestive evidence presented that, at least for the ImageNet dataset, popular classification models show substantial overfitting. An empirical example and perspectives from statistics highlight how models' (2) distribution generalization can benefit from consideration of causal relationships and counterfactual scenarios. Transfer learning approaches and results for (3) domain generalization are summarized, as is the wealth of domain generalization benchmark datasets available. Recent breakthroughs surveyed in (4) task generalization include few-shot meta-learning approaches and the emergence of transformer-based foundation models such as those used for language processing. Studies performing (5) modality generalization are reviewed, including those that integrate image and text data and that apply a biologically-inspired network across olfactory, visual, and auditory modalities. Higher-level (6) scope generalization results are surveyed, including graph-based approaches to represent symbolic knowledge in networks and attribution strategies for improving networks' explainability. Additionally, concepts from neuroscience are discussed on the modular architecture of brains and the steps by which dopamine-driven conditioning leads to abstract thinking.

cs.LG↗

arXiv2023

Problem-dependent attention and effort in neural networks with applications to image resolution and model selection

This paper introduces two new ensemble-based methods to reduce the data and computation costs of image classification. They can be used with any set of classifiers and do not require additional training. In the first approach, data usage is reduced by only analyzing a full-sized image if the model has low confidence in classifying a low-resolution pixelated version. When applied on the best performing classifiers considered here, data usage is reduced by 61.2% on MNIST, 69.6% on KMNIST, 56.3% on FashionMNIST, 84.6% on SVHN, 40.6% on ImageNet, and 27.6% on ImageNet-V2, all with a less than 5% reduction in accuracy. However, for CIFAR-10, the pixelated data are not particularly informative, and the ensemble approach increases data usage while reducing accuracy. In the second approach, compute costs are reduced by only using a complex model if a simpler model has low confidence in its classification. Computation cost is reduced by 82.1% on MNIST, 47.6% on KMNIST, 72.3% on FashionMNIST, 86.9% on SVHN, 89.2% on ImageNet, and 81.5% on ImageNet-V2, all with a less than 5% reduction in accuracy; for CIFAR-10 the corresponding improvements are smaller at 13.5%. When cost is not an object, choosing the projection from the most confident model for each observation increases validation accuracy to 81.0% from 79.3% for ImageNet and to 69.4% from 67.5% for ImageNet-V2.

cs.CV↗

arXiv2022

Forbidden Knowledge and Specialized Training: A Versatile Solution for the Two Main Sources of Overfitting in Linear Regression

Overfitting in linear regression is broken down into two main causes. First, the formula for the estimator includes 'forbidden knowledge' about training observations' residuals, and it loses this advantage when deployed out-of-sample. Second, the estimator has 'specialized training' that makes it particularly capable of explaining movements in the predictors that are idiosyncratic to the training sample. An out-of-sample counterpart is introduced to the popular 'leverage' measure of training observations' importance. A new method is proposed to forecast out-of-sample fit at the time of deployment, when the values for the predictors are known but the true outcome variable is not. In Monte Carlo simulations and in an empirical application using MRI brain scans, the proposed estimator performs comparably to Predicted Residual Error Sum of Squares (PRESS) for the average out-of-sample case and unlike PRESS, also performs consistently across different test samples, even those that differ substantially from the training set.

stat.ME↗

arXiv2022

A descriptive analysis of olfactory sensation and memory in Drosophila and its relation to artificial neural networks

This article provides a background and descriptive analysis of insect memory and the coding of olfactory sensation in Drosophila, presenting graphs and summary statistics from a large dataset of neurons and synapses that was recently made publicly available and also discussing findings from the existing empirical literature. Some general principles from Drosophila olfaction are discussed as they apply to the design of analogous systems in artificial neural networks: (1) the networks used for coding are shallow; (2) the level of connectedness varies widely across neurons in the same layer; (3) much communication is between neurons in the same layer; (4) in most olfactory learning, the manner in which sensory inputs are represented in stored memory is largely fixed, and the learning process involves developing positive or negative associations with existing categories of inputs.

q-bio.NC↗