 Name | DEEP COGNITION - Deep Learning Studio | SONY - Neural Network Console | IBM - Neural Network Modeler | NVIDIA - Digit | Comment |
|---|---|---|---|---|---|
| Version of the software | 2.0 | Current | 1.1 | 6.0 | For NNC and NNM it is the version currently available on their Cloud. |
| Underlaying Framework | Apache MXNet | SONY Neural Netwok Libraries | Theano Tensorflow/Keras Caffe (v1) | Tensorflow | |
| Development Activity | Important (2 enhanced versions have been release recently) | Unknown NNL - github is here : github.com/..." | Unkown | Important | NNC - Activity on the underlaying framework (NNL) is weak. |
| GPUs | CUDA compatible | ??? | Depends on the target Framework | CUDA compatible | |
| DL MODEL EDITOR | |||||
| Graphical editor provided | Yes | Yes | Yes | No Changing the model is possible by editing the network graph" | |
| DL models supported | |||||
| - CNN | Yes | Yes | Yes | Yes | |
| - LSTM | Yes | Yes | Yes | ||
| - MLP / Auto-encoders | Yes | Yes | Yes | ||
| - RNN | Yes | Yes | |||
| Pretained Networks | Yes (Resnet50, InceptionV3, SqueezeNet, VGG16, VGG19, WideResNet) | No | No | Yes | |
| Use Keras API | Yes | No | No | ||
| Save Keras HDFS file (.h5) | Yes | No | |||
| Save Keras model .yaml file | No | No | |||
| Save Keras model file | No | No | |||
| Load Keras model .yaml file | Yes | No | |||
| Load Keras HDFS file (.h5) | Yes | No | |||
| Code generation | No | No | Yes | ||
| - Theano | No | No | Yes | ||
| - Tensorflow/Keras | No | No | Yes | ||
| - Caffe (v1) | No | No | Yes | ||
| Custom cost function | Yes | No | No | ||
| Load Keras Models + Weights | Yes | No | |||
| Load Caffe Models + Weights | No | No | Yes | ||
| Load Keras Code | ??? | ||||
| Model checker | Yes | Yes | Yes | No | |
| Datasets Management | Yes | Yes | No | Yes | |
| Experiments Mangement | Yes | No | |||
| - Exp comparison | Yes | ||||
| - Prev. Exp loader | Yes | ??? | N/A | ||
| TRAINING | |||||
| DL model training | Yes | Yes | No | Yes | |
| Tune learning parameter | Yes | Yes | |||
| Live performance monitoring | Yes | No | |||
| INFERENCE | |||||
| Inference available | Yes | Yes | No | Yes | |
| Results browser | Yes | No | |||
| View Valid/test set inference | Yes | Yes | |||
| Confusion matrix | No | Yes | |||
| Visualize intermediate layer activation | Yes | No | |||
| User test int | Yes | No | |||
| DEPLOIEMENT | |||||
| - Rest API | Yes | Yes On DC server | |||
| PLATFORM | |||||
| - Cloud | Yes | Yes | |||
| but no computing environnement" | Yes | ||||
| - Cloud Trial period | Yes | Yes | Yes | ||
| - Desktop | Yes (any platform with python) | No | Yes | ||
| OTHER | |||||
| AutoML | Yes | No | No | No | |
| Browser | Yes | No | |||
| Environment management | Yes | No | No | ||
| Python shell | Yes | No | No | ||
| Jupyter notebook | Yes | No | No | ||
| Documentation | "Minimal | ||||
| (Keras API)" | "Minimal | ||||
| (SONY Neural Network Library)" | ??? | ||||
| Tutorials | Yes | Yes | Yes | ||
| Web Site | deepcognition.ai | sony.com/... | mybluemix.net/... | nvidia.com/... |