A state of the art Weighted and Projected Sampler
-
Updated
Jun 21, 2022 - Python
A state of the art Weighted and Projected Sampler
A digit bin index is a compact representation of a number of weights in the interval (0,1) structured for fast selection and removal of a particular weight based on the accumulated values of the contained weights.
1st place solution for the CentraleSupΓ©lec Deep Learning Kaggle challenge "3-MD-4040 2026 ZooCAM Challenge". Plankton image classification (1.2M samples, 86 classes) using CNNs trained from scratch and a weighted logits ensemble of ResNet50, EfficientNet-B3 and ConvNeXt-Tiny with label smoothing, weighted sampling and TTA.
Weighted sampling without replacement (Efraimidis-Spirakis A-Res and A-ExpJ) and reservoir sampling over streams, in pure Python with zero dependencies.
Proyecto que compara estimaciones de grupo usando diferentes tipos de muestreo. Algunas tΓ©cnicas de muestreo, los individuos de la poblaciΓ³n tienen diferentes probabilidades de ser seleccionados, lo cual es considerado por los modelos de estimaciΓ³n.
Bayesian network using weighted and Gibbs sampling
Algorithm R, Algorithm L, A-Res and Poisson samplers audited against exact first-order inclusion probabilities, chi-square uniformity and measured RNG-call cost
Pytorch model for estimating human age from facial images.
Cryptographically safe random utilities for Go. Helpers for generating random hex strings, tokens, numbers, bytes, UUIDs, emails etc. Thin wrappers around crypto/rand for everyday use.
A weighted random item sampler (selector), where the probability of selecting an item is proportional to its weight, and every item is sampled exactly once (without repetition or replacement). The sampling method utilizes a binary-search optimization, making it suitable for performance-demanding applications where the set of items is large.
A weighted random item sampler (selector), where the probability of selecting an item is proportional to its weight. The sampling method utilizes a binary search optimization, making it suitable for performance-demanding applications where the set of items is large and the sampling frequency is high.
Weighted random-sampling helpers β Vose's alias method, cumulative-bisect, Efraimidis-Spirakis A-Res. Zero dependencies.
NLP project: Markov Chain text generator with N-gram models and fast parallel preprocessing
To associate your repository with the weighted-sampling topic, visit your repo's landing page and select "manage topics."