Singapore makes mark on AI with Apache Singa

Apache Singa, the Singapore-developed distributed computing platform for training deep learning models over large datasets, has been conferred top-level project (TLP) status under the Apache Software Foundation.

Conceived by the database system research group at the National University of Singapore (NUS) School of Computing together with China’s Zhejiang University and online gaming company NetEase, the open-source platform had been an Apache incubator project since 2015.

Supported by Singapore’s National Research Foundation, Ministry of Education and Agency for Science, Technology and Research, Apache Singa partitions deep learning models and data onto cluster nodes to train the models.

“We saw an increasing demand for deep learning and machine platforms in 2012, but there was a lack of efficient distributed platforms,” said Ooi Beng Chin, the NUS professor that led the project.

Noting that the TLP status is a mark of recognition for Apache Singa, Ooi hopes the project “can make an impact on deep learning the same way Apache HTTP servers did for website servers”.

With its focus on healthcare applications and scalable architecture that can run on a range of hardware including graphics processing units (GPUs), Apache Singa is already being used by National University Hospital (NUH) and Singapore General Hospital to analyse MRI and X-ray images to better identify health problems.

NUH also uses models trained on Apache Singa to model the progression of diseases and patient re-admission trends.

Outside healthcare, Apache Singa is being used by SecureAge, a Singapore-based cyber security firm, to develop deep learning models for malware detection, in a bid to improve detection accuracy and identify new strains of malware using past data. 

In the financial sector, local banks are also using the deep learning platform to develop and train models for risk modelling and preventing money laundering. 

Apache Singa is the first project in Southeast Asia that has achieved TLP status, joining 300 other TLPs such as the Apache web server and Apache Kafka, the distributed data streaming platform designed to replace traditional message brokers.

Plans are afoot to make Apache Singa more accessible to people who are not trained in artificial intelligence, as well as to broaden hardware support to include edge devices where more inferencing than training currently takes place.

According to a new Github report, the number of open source contributors in Singapore grew by 111% over the past 12 months, the second-highest growth rate in Asia behind Hong Kong. Asia’s contributor community has also surpassed those in Europe and North America in annual growth.