AIA Analytics Engine

Capabilities

The AIA Analytics Engine is the technology that enables AIA Science to implement customer-specific artificial intelligence, machine learning and data analytics solutions, including:

feature-icon-1.svg

Prediction

Predict variables of interest such as the future price of a financial instrument or the energy consumption of a building where accurate measurements are not yet available.

feature-icon-2.svg

Classification

Classify input data into relevant categories. For example, which categories such as sports, politics, finance among others does a news article belong to.

feature-icon-3.svg

Recommendation

A truly scalable technology platform built with our High Performance Computing expertise.

feature-icon4.svg

Outlier and pattern detection

Detection of abnormal behavior that is useful for fraud detection or for flagging customers that are about to churn.

feature-icon-5.svg

Clustering

Find patterns in a data set. For example grouping similar customers together to distribute among sales representatives, identifying groups of similar leads, or finding clusters of similar documents (without necessarily considering the corresponding topics).

feature-icon-6.svg

Video/Image analytics

Analyse video and image data to detect objects such as recognizing bicycles and cars going through a live video stream or using the content of an image to extract relevant features that are useful for prediction and/or classification tasks.

feature-icon-7.svg

Network/Graph analytics

Leverage the many-to-many relationships that exists in certain applications, such as those involving social network data, to generate insight that would otherwise be hard to find.

Scalability

AIA Science have in-house expertise and many years of experience on both scale-up systems (powerful servers) and scale-out systems (cluster of servers running in parallel). We use our evaluation suite to benchmark each individual application and propose the optimal balance between scale-up and scale-out depending on the load parameters of the system. We offer deployment on our engine either using on-premise dedicated server or in our cloud solution.

Hardware Optimizations

section-icon-2.svg

AIA Science is a spinout of NUMASCALE, a pioneer in High Performance Computing (HPC). Numascale delivers true shared-memory programming and simpler administration at standard HPC cluster price points. In addition to this ongoing collaboration with NUMASCALE, AIA Science inherited their software IP, named here AIA Acceleration

AIA ACCELERATION INCLUDES:

  • Optimized turnkey Linux installation (bootloader, kernel, OS) in both clusters and Scale-up systems
  • System libraries for doing NUMA-aware memory allocations
  • Optimized linear algebra libraries
  • Optimized functions in the programming language R

Linear algebra is the backbone behind all essential Machine Learning algorithms, meaning that AIA Acceleration allows us to generate faster predictions for our customers.

AIA Acceleration and AIA Science Engine can easily be integrated with customer-specific platforms, and aligns well with other technologies such as Microsoft Azure, AWS and Google Cloud, as well as SAP HANA.

Data Pipelines

We leverage cutting-edge open source software, such as Apache Kafka, Apache Spark, Apache Hadoop, among others, to ensure that all the statistical models and relevant system information are always up-to-date with respect to new data that continuously flows into the system.

AIA Analytics of Things

AIA Analytics of Things (AoT) is a framework for IoT devices and communication that turns the AIA Analytics Engine into a scalable IoT platform, cutting months of development time for companies that want to go digital.

It is scalable and capable of handling anything from moderate to extremely large amounts of data, and delivers near real-time analytics capabilities to ensure that you get the maximum performance out of your assets and can react ahead of your competitors and avoid major negative incidents.

In sum, a modern, powerful
Data Analytics Engine

We are continually working to make sure that the state of the art methods and technologies in machine learning and artificial intelligence are made available through our analytics Engine, including generalized additive models (GAM), times-series models such as autoregressive and moving-average models (ARMA) and deep learning models such as convolutional neural networks (CNN) and recurrent neural networks (RNN), to name a few.

AIA Analytics allows customer code to be written in R and Python and, using highly optimized proprietary libraries, being executed in parallel on an advanced scale-up architecture. Many R implementations are not able to handle very large datasets and are too slow for application areas requiring near real-time updates.

AIA Science solves this problem by enabling full utilization of all cores in large scale-up systems and by providing the largest amounts of memory possible to ensure that even the most complex calculations can be performed without running out of resources.

The AIA Science optimized hardware uses an integrated hardware platform and advanced math libraries to tackle the most challenging computing workloads in the industry. This results in fast and scalable execution of your analytics code and models – without demanding a real change in how your work.

connect with AIA Science

Have a challenging project to discuss?

The problem is not to find technology, but to use it the right way. If you have already invested in a cloud platform, our expertise, methodology and architecture frameworks can help accelerate your time to market, and ensure that you end up with an enterprise solution that keeps up as your data and business changes.

AIA_logo.svg

Copyright 2017 - AIA Science