m
Our Mission Statement
This is Photoshop's version of Loremer Ipsn gravida nibh vel velit auctoregorie sam alquet.Aenean sollicitudin, lorem quis bibendum auci elit consequat ipsutis sem nibh id elit.
Follow Us
Top
Machine Learning - VoltDB
12974
page-template-default,page,page-id-12974,page-child,parent-pageid-12986,cookies-not-set,mkd-core-1.0,highrise-ver-1.0,,mkd-smooth-page-transitions,mkd-ajax,mkd-grid-1300,mkd-blog-installed,mkd-header-standard,mkd-sticky-header-on-scroll-up,mkd-default-mobile-header,mkd-sticky-up-mobile-header,mkd-dropdown-slide-from-bottom,mkd-dark-header,mkd-header-style-on-scroll,mkd-full-width-wide-menu,mkd-header-standard-in-grid-shadow-disable,mkd-search-dropdown,mkd-side-menu-slide-from-right,wpb-js-composer js-comp-ver-5.4.7,vc_responsive
 

Machine Learning

VoltDB / Solutions / Machine Learning

Machine Learning

Make intelligent decisions in real-time, when it matters most.

Legacy database technology was focused on analyzing historical data to gain an rear-view understanding of business performance. While it is important to analyze where the business is coming from, in order to gain the competitive advantage and differentiate your application it is critical to utilize deep learning and take action in-event; with the end goal of driving desirable business outcomes.

Building a robust predictive analytics model is only half the battle. Utilizing the model in production for real-time decision making is the key element of Machine Learning, this however is easier said than done. Most organizations have historical data stored in multiple places such as a data warehouse, data lake, ERP system, and more. In addition to existing data, a high volume of data is constantly streaming in from multiple sources at a very high velocity. In an production environment, the model needs to continually ingest, train on historical data and operationalize in real-time at very low latency.

VoltDB was designed for operationalizing Machine Learning in real-time. VoltDB can seamlessly:

  • Ingest and analyze fast streaming big data while simultaneously querying historical data to continuously improve the model.
  • Import complex models built by data scientists through PMML, PFA, and other standards ensuring quick implementation and consistency.
  • Process incoming events in real-time based on complex Machine Learning rules.
  • Offer unparalleled high performance with sub millisecond response times. Concurrently update the models in hot-path while maintaining system availability.
  • Deliver service dynamically leading to high agility and reduced total cost of ownership.
  • Guarantee ACID compliance, which is especially important for Enterprise applications in financial services and telecommunications.

Use Cases for Real-time Machine Learning

Fraud Detection

By changing your fraud system from a post-transaction detection system to a proper in-transaction prevention system, you can lower operating costs, reduce false positives, and stop fraud as it happens.

Hyper-personalization

A real-time application or service must predictively and actively engage customers with highly personalized experiences. Achieving this with fast data requires tools that can collect, explore, analyze, and act on multiple streams of data instantaneously. These tools allow businesses to make data-driven decisions using insights from real- time analytics against fast moving data.

Huawei Selects VoltDB to Power their Financial Services Fraud Detection Solution FusionInsight

The Challenges:

  • Move fraud detection from “post” transaction to in-line.
  • Expand the FusionInsight platform with live real-time fraud analysis.
  • Find a system that can be easily deployed in FusionInsight, that offers: deployment flexibility, manageable cost, and financial-grade security.

The Solution:

  • VoltDB embedded in FusionInsights could perform:
  • Thousands of queries per financial transaction with low latency (<50ms) and high throughput (>10k tps).
  • Monitor 10,000 complex transactions per second with 99.99% of transactions finishing in less than 50ms.
  • Apply hundreds of rules and scoring checks to each transaction in milliseconds, moving fraud detection from weeks to real-time.

The Results:

  • 10x better performance than traditional fraud detection systems.
  • More than a 50% reduction in fraud cases.
  • Over $15M/year savings from fraud loss prevention.

Read the full Huawei case study now.