Xelera Silva

Xelera Silva

Gradient Boosting frameworks such as XGBoost, LightGBM and CatBoost, as well as Random Forest algorithms are widely used in ransomware and DDOS detection systems, recommender systems and trading systems. Xelera Silva a Decision Tree acceleration software provides best-in-class throughput and latency for XGBoost, LightGBM and Random Forest inference by leveraging COTS datacenter-grade FPGA accelerators.

Key benefits

• Acceleration: 10x higher throughput than GPUs, 50x higher throughput than CPUs
• Guaranteed latency: Milliseconds query latency with 0.1ms jitter
• Seamless adoption: Usable with standard machine learning frameworks in a bring-your-own-model fashion, usable with zero code change
• Deployment: Built to run out-of-the-box on COTS server and datacenter FPGA hardware, and on public cloud instances

Benchmark: Kaggle Malware Detection dataset (https://www.kaggle.com/c/malware-detection/overview) |Number of features: 23 | Maximum decision tree depth: 10 levels | Number of decision trees: 200

Example use-cases

• Network security (DDOS prevention)
• Malware detection
• Anomaly detection
• Recommendation engines
• Risk management
• Algorithmic trading

Application areas

Network Threat Detection

The network is the first instance that must be protected in order to secure the enterprise IT system. Machine learning-powered firewalls perform application-level inspection to detect suspicious patterns in the network flow data (e.g. in the case of a DDOS attack). Silva provides the inferencing speed required to ensure that computationally demanding machine learning-based firewalls can operate at an adequate speed to avoid a slowing-down of the IT system.

Ransomware Detection

Ransomware cyber attacks are rising at a rapid pace and traditional security approaches often fail to intercept them. Gradient Boosting Decision Tree and Random Forest algorithms are among the most widely used techniques for the automated detection of ransomware attacks. A high filtering throughput and low detection latency are the key objectives in these systems. Silva satisfies these performance constraints by ensuring a high throughput at an ultra-low latency.

Risk Management

The value at risk assessment and the credit risk assessment of contract counterparties are two statistical techniques to manage the risk in financial transactions. Both techniques shift more and more frequently towards real-time processes. On the other hand, both techniques often rely on data- and compute-intensive Monte Carlo methods. Xelera Analytics makes these methods real-time capable.

Algorithmic Trading

Algorithmic trading usually relies on time series analyses using Markov-like model such as autoregressive or moving average models, while Deep Learning-based techniques are under investigation. The more complex these models are, the more compute delay they cause. Xelera Analytics help keep this compute delay at a minimum, which is particularly important in high- frequency trading.

Availability

• Public Cloud (AWS, Nimbix)
• On-premises
• Edge Clouds

Demo License Request

English:

Xelera Technologies GmbH collects, processes, and uses your personal data above to process your request.

Possibility of revocation:

Of course, you can revoke your consent at any time with effect for the future. You can revoke your consent at any time in the future by sending an e-mail to info@xelera.io or by contacting Xelera Technologies GmbH using the contact information provided in the legal notice. Your revocation does not affect the lawfulness of the processing of your data that took place on the basis of your consent until the revocation. You can find further explanations on data protection at Xelera Technologies GmbH at https://xelera.io/privacy-policy.

German:

Die Xelera Technologies GmbH erhebt, verarbeitet und nutzt Ihre vorstehenden personenbezogenen Daten zur Bearbeitung Ihres Anliegens.

Widerrufsmoeglichkeit:

Selbstverstaendlich koennen Sie Ihre Einwilligung jederzeit mit Wirkung fuer die Zukunft widerrufen. Den Widerruf koennen Sie gegenueber der Xelera Technologies GmbH ueber folgende E-Mail geltend machen: info@xelera.io bzw. die Xelera Technologies GmbH hierzu unter den im Legal Notice angegebenen Kontaktdaten kontaktieren. Ihr Widerruf laesst die Rechtmaessigkeit der aufgrund der Einwilligung bis zum Widerruf erfolgten Verarbeitung Ihrer Daten unberuehrt.

Weitere Erlaeuterungen zum Datenschutz bei der Xelera Technologies GmbH finden Sie unter https://xelera.io/privacy-policy.