DCNNGasClassification-Python

DCNN分类电子鼻数据集，并尝试使用SDA进行漂移补偿（效果不好）

Ref:

1. An optimized Deep Convolutional Neural Network for dendrobium classification based on electronic nose

2. Gas Classification Using Deep Convolutional Neural Networks

3. Domain Adaptation for Large-Scale Sentiment Classification: A Deep Learning Approach

4. ArcFace：Additive Angular Margin Loss for Deep Face Recognition

Datasets:

1. 10 boards, Ref: Chmical gas sensor drift compensation using calssifier ensembles
2. 5 boards, Ref: Calibration transfer and drift counteraction in chemical sensor arrays using Direct Standardization

错误（已修改）:

在network2中，BN层之前的conv层不应使用激活函数。ReLU激活函数应放置在BN层之后。(并未明显影响精度)