TigerGraph: Machine Learning Powered by Graph Analytics

The new kid on the block in machine learning is TigerGraph ML Workbench, a robust toolbox helping data scientists to increase ML model correctness, shorten development cycles, and value-add to the company.

CIO World Asia conducted a deep dive with Victor Lee, Vice President of Machine Learning and Chung Ho, GM for Asia Pacific to find out more about graph-enhanced machine learning (ML).

---

Elevate Businesses With Machine Learning

Increasing potential revenue while decreasing cost and risk is at the heart of businesses. ML enables enterprises to reach deeper into this goal. Making predictions is ML’s key function. The best predictions have high accuracy scores and real-world application. With better predictions, businesses can identify optimal opportunities in making the best recommendations to customers, matching them with products they actually would consider purchasing. 

E-commerce personalised product recommendations are based on the website’s knowledge of its users. Websites have no past information on first time customers, yet it is still possible to categorise newcomers and offer personalised recommendations. This is done by grouping newcomers with customers of similar characteristics — IP address location, time of day of access, accessing platform (web or mobile). Past data on past success (i.e. purchases), are gathered to match the present pattern in the newcomer’s browsing activity, to make an educated guess on ideal product recommendations. On the other hand, returning customers’ recommendations are informed by their browsing history, orders, returns and exchanges. The accumulation of data in turn, snowballs to compute, analyse, improve future recommendations.

Graph analytics and machine learning is helping businesses to meet their goals better by making better predictions, increasing revenue or cutting costs. Machine learning helps companies do what they do better, by analysing data more deeply, thereby making more accurate predictions.

Victor Lee, Vice President of Machine Learning

Benefits And Challenges Of Improving Machine Learning

Improvements in ML are often incremental. Advancements allow ML to make recommendations and detect fraud better than before. Sometimes a completely new innovation enters the sphere. In the likes of human language recognition, ML is trained to independently listen to and answer questions from humans. 

Planning the transition phases is key to successfully implementing new technology. Outlining the new technology’s integration into existing systems, ensuring its reliability and scalability. Some products and services are ready to use out of the box, but not TigerGraph. TigerGraph is a general purpose capability platform to build ML applications. The product is targeted at enterprises with intelligence and resources to train the application to their desired purposes.

TigerGraph ML Workbench’s Game-Changing Capabilities

TigerGraph boosts a large customer base in financial services. Fraud detection software is high on the agenda of fintech professionals. Customers use TigerGraph’s analytics with their  existing fraud detection systems for high detection rates. There are 2 types of financial fraud – transaction and account fraud. Fraudulent transactions revolve around suspicion that credit card payments are not made or authorised by the real cardholder. Account fraud detection comes into play when banks conduct background checks on clients requesting to open accounts. Before predicative software entered the scene, all these were guesswork. 

Overseas credit card usage 10 years ago mandated a call to the card issuer, requesting for overseas activation and providing a heap of personal information to prove one is the real cardholder. Otherwise, the card may be frozen for suspected fraudulent activity. With ML intelligence, applications are able to make educated inferences, connecting air ticket purchases to a card user’s plausible overseas travel, thus not suspecting these transactions.

Gathering more information and applying graph algorithms work hand-in-hand to link the individual user to a greater community with highly interconnected things. Rapid graphical analysis across multiple huge datasets with the end result of giving a probability score is what TigerGraph brings to the table. 

TigerGraph’s Pipeline Growth

ASEAN has witnessed the double digit growth of TigerGraph in her region. The doubled customer base continues the sharp growth trend since the onset of the pandemic. The company’s technology is especially up-taken by global supply chain enterprises hungry for insights to survive tough times. These enterprises are weary of using traditional technology to unearth strategic insights. TigerGraph’s graphical approach provides a fresh way of perceiving and interpreting data. As Gartner forecasted, 80% of data and analytics innovations will be done using graph technology by 2025. ML-enabled graphical analysis has sights to be a continued pillar of growth. 

---