Download JAR files tagged by only with all dependencies

A platform for discovery and communication between various parts of a distributed service or application. At its minimum, the platform is a library which can be used to serialize and deserialize application constructs into payloads which it can then send to and receive from remote endpoints. For now it supports TCP and HTTP communication protocols with WebSockets coming soon. It supports raw byte streams via TCP, as well as application/json via JAX-RS as serialization protocols. It offers plain or SSL/TLS communication for security of your data in transit. For now it is only available in Java, with bindings for other languages coming soon. The communication pattern is request-reply, with fail fast semantics, and with service consumers (clients) knowing beforehand the coordinates of the service providers (servers). The TCP communication protocol offers blocking/unblocking/asynchronous flavors whereas the rest are only blocking for now. Coming soon, the platform will provide a Discovery Service to facilitate the registration of the service providers (servers) and service consumers (clients). In this case, the callers don't need to resolve or even balance their calls towards various endpoints, since the Platform handles this tasks. In this case, the pub/sub communication pattern will also be available (coming soon) with at-most-once delivery guarantees. We intend to provide a Security Service for managing the keys/and certificates of various services as well as Metrics and Logging services (coming soon).

This package currently contains classes for training multilayer perceptrons with one hidden layer, where the number of hidden units is user specified. MLPClassifier can be used for classification problems and MLPRegressor is the corresponding class for numeric prediction tasks. The former has as many output units as there are classes, the latter only one output unit. Both minimise a penalised squared error with a quadratic penalty on the (non-bias) weights, i.e., they implement "weight decay", where this penalised error is averaged over all training instances. The size of the penalty can be determined by the user by modifying the "ridge" parameter to control overfitting. The sum of squared weights is multiplied by this parameter before added to the squared error. Both classes use BFGS optimisation by default to find parameters that correspond to a local minimum of the error function. but optionally conjugated gradient descent is available, which can be faster for problems with many parameters. Logistic functions are used as the activation functions for all units apart from the output unit in MLPRegressor, which employs the identity function. Input attributes are standardised to zero mean and unit variance. MLPRegressor also rescales the target attribute (i.e., "class") using standardisation. All network parameters are initialised with small normally distributed random values.