Understanding the Effective Receptive Field in Deep Convolutional Neural Networks

原創 DreamLike_zzg 2020-06-24 13:36

相關參考
https://zhuanlan.zhihu.com/p/86599013

ERF比理論感受野小
近似成高斯分佈
subsampling、diation conv 可增大ERF
訓練過程中ERF不斷增大

 

發表評論
所有評論
還沒有人評論，想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章

2020-07-03 CVPR2020 i3DV論文討論（2） 筆記

目錄[1] Context Prior for Scene Segmentation[2] Deep Stereo using Adaptive Thin Volume Representation with Uncertaint

NeverMoreH 2020-07-04 09:51:10

2020-07-06 組會 筆記

目錄賈老師 賈老師 利他主義很重要 學習狀態，自律 清華兩篇博士論文，有思想，有深度 優博名額5%->10%，今年有2.5個名額

NeverMoreH 2020-07-06 14:22:03

2020-07-02 CVPR2020 V&L論文討論（3） 筆記

目錄[1] Bi-directional Relationship Inferring Network for Referring Image Segmentation[2] A Real-Time Cross-modality

NeverMoreH 2020-07-04 10:30:59

2020-06-18 CVPR2020 V&L論文討論（1） 筆記

目錄CC JJY S[1] Normalized and Geometry-Aware Self-Attention Network for Image Captioning[2] Cops-Ref: A new Dataset

NeverMoreH 2020-07-04 10:30:59

2020-07-01 CVPR2020 表示學習論文討論（2） 筆記

目錄[1] CVPR2020_A Unified Optimization Framework for Low-Rank Inducing Penaltie[2] CVPR2020_Automatic Neural Network

NeverMoreH 2020-07-04 09:51:10

Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles

普通的機器學習分類迴歸問題都是點估計，即模型給出的輸出是一個real value，或者是各個類別的probability。 但是模型對於不同點的估計，確信度應該是不同的。對於和以往出現過樣本非常相似的點，給出的預測確信度比較高，對於和以往

nick__huang 2020-07-02 09:52:41

Paper Reading: WGAN & WGAN-gp

Paper Reading Note URL: Towards Principled Methods for Training Generative Adversarial Networks Wasserstein GAN T

waple_0820 2020-06-29 07:25:18

Adaptive Cross-Modal Few-shot Learning

論文 資料1 基於度量的元學習（metric-based meta-learning）如今已成爲少樣本學習研究過程中被廣泛應用的一個範式。這篇文章提出利用交叉模態信息（cross-modal information）來進一步加強現有的度量

vieo 2020-06-28 22:43:10

Cross Attention Network for Few-shot Classification

作者：一顆檸檬味的橙子 鏈接：https://zhuanlan.zhihu.com/p/105717426 來源：知乎 著作權歸作者所有。商業轉載請聯繫作者獲得授權，非商業轉載請註明出處。   來源：NeurIPS 2019 文章題目：C

vieo 2020-06-28 22:43:10

[paper] GAN

(NIPS 2014) Generative Adversarial Nets Paper: https://papers.nips.cc/paper/5423-generative-adversarial-nets Code

lijiancheng0614 2020-06-24 21:32:37

[paper] Look Into Person

(CVPR 2017) Look into Person: Self-supervised Structure-sensitive Learning and A New Benchmark for Human Parsing P

lijiancheng0614 2020-06-24 21:32:37

[paper] Multiple Human Parsing

Towards Real World Human Parsing: Multiple-Human Parsing in the Wild Paper: https://arxiv.org/pdf/1705.07206.pdf

lijiancheng0614 2020-06-24 21:32:37

[paper] DCGAN

(ICLR 2016) Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks Paper: ht

lijiancheng0614 2020-06-24 21:32:37

[paper] Learning to Remember Rare Events

(ICLR 2017) Learning to Remember Rare Events Paper: https://openreview.net/pdf?id=SJTQLdqlg Code: https://github.

lijiancheng0614 2020-06-24 21:32:37

輕量化backbone

MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications  (2017.04) 1、使用 depthwise conv + p

DreamLike_zzg 2020-06-24 13:36:17