yinglang / Cifar10_mxnet
使用mxnet编写的kaggle CIFAR10比赛的代码
Stars: ✭ 68
Labels
Projects that are alternatives of or similar to Cifar10 mxnet
Lovaszsoftmax
Code for the Lovász-Softmax loss (CVPR 2018)
Stars: ✭ 1,148 (+1588.24%)
Mutual labels: jupyter-notebook
Puzzlemix
Official PyTorch implementation of "Puzzle Mix: Exploiting Saliency and Local Statistics for Optimal Mixup" (ICML'20)
Stars: ✭ 67 (-1.47%)
Mutual labels: jupyter-notebook
Encode Attend Navigate
Learning Heuristics for the TSP by Policy Gradient
Stars: ✭ 68 (+0%)
Mutual labels: jupyter-notebook
Twitter sentiment analysis
A guide for binary class sentiment analysis of tweets.
Stars: ✭ 68 (+0%)
Mutual labels: jupyter-notebook
Data bootcamp
Materials for a course at NYU Stern using Python to study economic and financial data.
Stars: ✭ 67 (-1.47%)
Mutual labels: jupyter-notebook
Covid 19 Dataviz
Simple data visualization on Covid-19 data using Pandas and Google Colaboratory
Stars: ✭ 68 (+0%)
Mutual labels: jupyter-notebook
Equivariant Transformers
Equivariant Transformer (ET) layers are image-to-image mappings that incorporate prior knowledge on invariances with respect to continuous transformations groups (ICML 2019). Paper: https://arxiv.org/abs/1901.11399
Stars: ✭ 68 (+0%)
Mutual labels: jupyter-notebook
Backdrop
Implementation and demonstration of backdrop in pytorch. Code and demonstration of GP dataset generator.
Stars: ✭ 68 (+0%)
Mutual labels: jupyter-notebook
Qpga
Simulations of photonic quantum programmable gate arrays
Stars: ✭ 68 (+0%)
Mutual labels: jupyter-notebook
Equalareacartogram
Converts a Shapefile, GeoJSON, or CSV to an equal area cartogram
Stars: ✭ 68 (+0%)
Mutual labels: jupyter-notebook
Predictive Maintenance
Demonstration of MapR for Industrial IoT
Stars: ✭ 68 (+0%)
Mutual labels: jupyter-notebook
Hass Amazon Rekognition
Home Assistant Object detection with Amazon Rekognition
Stars: ✭ 68 (+0%)
Mutual labels: jupyter-notebook
CIFAR10_mxnet
abstract
kaggle CIFAR10 compitition code, implement by mxnet gluon.
we got 0.9688 by merge some ideas from https://discuss.gluon.ai/t/topic/1545/423
directroy and file descriptor
| file | descriptor |
|---|---|
| log | some train log file |
| models | some trianed model params(weight) |
| result | the forward result file on kaggle test set |
| submission | the finnal kaggle submission result |
| CIFAR10_train | main train and exp code |
| plot | the visulization of train acc and valid acc and loss with epoch |
| utils/netlib.py | ResNet18, ResNet164_v2, densenet, Focal Loss implement code by gluon, invoke by CIFAR10_train |
| utils/utils.py | some tool function |
models, reuslt, log can get from link: https://pan.baidu.com/s/1pLjzQWj key: f6p3
method description
the main idea is from mxnet topic,we merge the most ideas. first, we train ResNet164_v2 in diffrent data argument policy sencondly, we use 'focal loss' replace 'softmax cross entropy loss' thirdly, we use densenet replace ResNet164_v2 lastly, we ensemble some net to get higher acc, we found this five models get best score:
| policy | kaggle score |
|---|---|
| res164_v2 + DA1 | 0.9529 |
| res164_v2 + DA2 | 0.9527 |
| res164_v2 + focal loss + DA3 | 0.9540 |
| res164_v2 + focal loss + DA3(only use 90% train_data) | 0.9506 |
| sherlock_densenet | 0.9539 |
DA1~DA3 is means diffrent data argument
| DA | policy |
|---|---|
| DA1 | padding image to 40, and then random crop (32, 32)，same as code in sherlock |
| DA2 | resize image to a bigger size，and then crop to (32, 32)，and set data argument parma of HSI to 0.3,PCA noise to 0.01. |
| DA3 | after DA2, clip the color of image to (0,1)（make the generate image more friendly to human） |
we ensemble the five net, and got 0.9688 score.
目录文件描述
| 文件名 | 描述 |
|---|---|
| log | 一些训练的日志，主要是训练的loss和acc |
| models | 一些训练的模型参数(权值) |
| result | 程序forward的最终结果，保留了10个类别的output |
| submission | 最终提交的结果文件 |
| CIFAR10_train | CIFAR10上训练模型和产生结果的代码，主程序。 |
| plot | 绘制一些模型训练过程的精度和loss曲线 |
| utils/netlib.py | ResNet18, ResNet164_v2, densenet, Focal Loss 的gluon的实现,被调用。 |
| utils/utils.py | 一些工具函数 |
models、result、log等内容加起来有点大，等传到网盘上，供各位小伙伴参考下载。 链接: https://pan.baidu.com/s/1pLjzQWj 密码: f6p3
方法描述
参考论坛几个小伙伴的方法，我总结了一下，大致如下: 使用不同的网络的数据增强的方法，我们做了多个实验，得到了多个网络模型(全部放到了models下面)，然后ensemble，发现下面5个网络的效果最好。 这5个网络的训练策略和单独提交的精度分别是：
| policy | kaggle 精度 |
|---|---|
| res164_v2 + DA1 | 0.9529 |
| res164_v2 + DA2 | 0.9527 |
| res164_v2 + focal loss + DA3 | 0.9540 |
| res164_v2 + focal loss + DA3 | 只使用train_data训练: 0.9506 |
| sherlock_densenet | 0.9539 |
上面的DA是3中不同的数据增强的方法:
| DA | policy |
|---|---|
| DA1 | 就是最常用的那种padding到40,然后crop的方法，就是sherlock代码里使用的加强 |
| DA2 | 是先resize到一定的大小，然后crop的方法，同时设置了HSI的几个参数为0.3,PCA噪声为0.01 |
| DA3 | 时在DA2后，将图片的颜色clip导（0,1）之间（动机时创建更符合人感官的自然图片数据） |
五个网络按照各自的精度加权求和作为最后的结果，就有了0.9688的效果。
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at [email protected].