Simpler, Faster, Stronger: Breaking The log-K Curse On Contrastive Learners With FlatNCE

FlatNCE: 避免浮点数误差的小批量对比学习损失函数.

CompConv: A Compact Convolution Module for Efficient Feature Learning

CompConv：使用分治法的紧凑卷积模块.

Integrating Circle Kernels into Convolutional Neural Networks

卷积神经网络中的圆形卷积核.

YOLOX: Exceeding YOLO Series in 2021

YOLOX：Anchor-free的YOLO检测器.

局部线性嵌入(Locally Linear Embedding, LLE)

Locally Linear Embedding.

等度量映射(Isometric Mapping, ISOMAP)

Isometric Mapping.