Cloudifier Virtual Apps – Virtual desktop predictive analytics apps environment based on GPU computing framework

The need for systems capable of conducting inferential analysis and predictive analytics is ubiquitous in a global information society. With the recent advances in the areas of predictive machine learning models and massive parallel computing a new set of resources is now potentially available for the computer science community in order to research and develop new truly intelligent and innovative applications. In our research we present the principles, architecture and current experimentation results for an online platform capable of both hosting and generating intelligent applications – applications with predictive analytics capabilities.

The entire field of Artificial Intelligence tends to become more and more about improving predictions and predictive models as Goldfarb et al argue in the paper “Managing the Machines”. Predictive modelling can now be used for extremely varied tasks ranging from constructing expert system chat-bots based on Long Short Term Memory deep recurrent neural networks that would use natural language processing capability (reading and writing) up to melanoma recognition in dermoscopy images based on deep convolutional networks.

We propose several approaches:

(i) a web-services based model for the deployment of model training and prediction on GPU massive parallel computing backend. The GPU based parallel computing environment will enable to construct and optimize a GPU-optimized set of machine learning algorithm implementations.

(ii) employment of automatic software design, development and deployment for user experience modules based on a pipeline shallow machine learning model.

(iii) a unified architecture and underlying framework for generation and hosting predictive analytics applications.

Our proposed research and experimentation will fuel the development and implementation of a production-grade system capable of addressing in the same time the needs of several categories of customers such as:

• users in need of a online application marketplace fully oriented on predictive analytics applications;
• software developers in need of a legacy apps automatic software translation tools including machine learning templates;
• scientific users both from academic and commercial environment looking for an online application framework and execution environment in order to conduct predictive experiments in areas such as healthcare, finance and environment;