DeepFlare is a multi-use python library, including algebraic functionallities, bivariate-statistics modules, deep-learning architectures, and tensor processing features. This library is constantly being built on, where as time progresses new modules are being introduced. In terms of effectiveness on Mnist the DeepFlare's feed forward neural network is reaching approximately:
- 0.92 after 1 epoch.
- 0.96 after 2 epochs.
- 0.97 after three epochs.
- 1.0 on training data after 20 - 25 epochs.
The accumulation of these factors evidently articulate the strength and potential of the model.
Deeflare is also recognized for other functionallities:
- Quadratic Regression
- Linear Regression
- Cubic Regression
- Correlation Coefficient
- Preprocessing tools
- Tensor Proceessing Modules
- Analysis Section
- (Soon to come: Transformer Architecture Modules)
- (Soon to come: AI analyis feature)
Deeflare demonstrates the perfect balance between light-weight simple tools, and powerful modules. And while DeeFlare is currently on pause, further development stages are in place, and being developed.