DIANNE v0.4.0

Features:

• Major revision of Tensor library resulting in serious performance gains
• No more Java Tensor implementation in favour of simpler code
• Change build system from ant to gradle
• Fixed issues with native compilation not being performed correctly, especially with cutorch
• Improved performance for minibatches
• Improved memory footprint by reusing tensors where possible & triggering garbage collection
• Added various new modules, a.o. FullConvolution, BatchNormalization, ...
• Added various new composite modules, a.o. Inception, Residual, ...
• Extended Convolution to include temporal & volumetric operation
• MaxPooling has indices as optional second output, which can be passed to new MaxUnpooling module
• More efficient loading of datasets
• Added descriptors for datasets, allowing to define custom datasets without coding
• Added dataset adapters for data augmentation, a.o. flip, rotate, crop, ...
• Separated evaluator into generic & classification
• Added more stop criteria for a learn job based on error
• Added options for learning rate decay in SGD
• Perform validation while learning
• Automatically save best weights based on validation error
• Better error handling for jobs (especially on NaN)
• Easier resubmission for jobs, incl. changing parameters
• Fixed issue with URLs redirecting indefinitely
• UI improvements, such as ability to remove modules, job-specific URLs, etc.
• Extended JSON RPC
• Clean up of run configurations
• Various bug fixes
• Updated documentation to reflect new features in more detail
• Added contributor license agreement

DIANNE v0.3.0

Features:

• Major refactoring of learn API
• Added coordinator service for centralized management of tasks (learn, evaluate, act)
• Added dashboard UI for easy submission and follow up of said tasks
• Added JSON RPC for external systems to submit tasks
• Neural networks as composite modules
• Support for recurrent neural nets with a memory module
• Util API for easy programmatic construction of networks
• Improved test setup
• Increased performance for gradient operations on GPU
• Changed build system to gradle
• Separated native libraries into separate bundle fragments
• Improved error handling
• Various small additions and bug fixes

DIANNE v0.2.0

Features:

• DTOs for representing neural networks and their modules
• NeuralNetworkPlatform API as central management API
• NeuralNetwork API facilitating forward/backward calls of neural network
• Processor API for setting up different Trainers
• Distributed training possible synced by Repository
• Reinforcement Learning APIs for modelling Environments, Agents, ...
• Deep Q Learning agent implementation
• Pong RL example
• ALE environment
• DianneCoordinator API for training nn models
• JSON RPC interface to DianneCoordinator
• Add open source license, basic documentation and website
• Various small additions and bug fixes

DIANNE v0.1.0

Initial Dianne Release

Supports

• Modular deployment of neural network
• Supports fully connected and convolutional layers
• Supports basic split/fork/join/concat modules
• Save/load neural network modules
• GUI neural network builder
• Tensor library implementations in plain Java, native based on TH (Torch), cuda lib based on THC
• Basic training procedure for simple neural networks