Learning Transferable Architectures for Scalable Image Recognition

Developing neural network image classification models often requires significant architecture engineering. In this paper, we study a method to learn the model architectures directly on the dataset of interest... (read more)

PDF Abstract CVPR 2018 PDF CVPR 2018 Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Image Classification ImageNet NASNET-A(6) Top 1 Accuracy 82.7% # 50
Top 5 Accuracy 96.2% # 28
Number of params 88.9M # 20
Image Classification ImageNet ReaL NASNet-A Mobile Accuracy 81.15% # 18
Image Classification ImageNet ReaL NASNet-A Large Accuracy 87.56% # 13

Methods used in the Paper


METHOD TYPE
Neural Architecture Search
Neural Architecture Search
Sigmoid Activation
Activation Functions
Tanh Activation
Activation Functions
Entropy Regularization
Regularization
PPO
Policy Gradient Methods
Exponential Decay
Learning Rate Schedules
Instance Normalization
Normalization
Layer Normalization
Normalization
Dropout
Regularization
RMSProp
Stochastic Optimization
Weight Decay
Regularization
SGD with Momentum
Stochastic Optimization
Batch Normalization
Normalization
1x1 Convolution
Convolutions
Convolution
Convolutions
ReLU
Activation Functions
Softmax
Output Functions
LSTM
Recurrent Neural Networks
ScheduledDropPath
Regularization