As AI technology spreads across the globe, new locations are arising as potential hotbeds for the growth and development of AI technology. One of these areas is Southeast Asia, where AI solutions are expected to lead to increased productivity and a further expansion of AI markets.
To learn more, we talked to Adam Gibson, the head of Skymind Global Ventures’ AI division, Konduit AI. SGV recently launched an $800 million fund to back new AI companies and academic research, and generate business opportunities across Europe and Asia. In this article we talk with Gibson about his work at Konduit AI and the potential for AI implementation in the SE Asia market.
Adam Gibson, Cofounder of Deeplearning4j and Skymind
Though he currently heads up AI as part of Konduit, Gibson is most well known for creating the Eclipse Deeplearning4j (DL4J) framework in 2013. DL4J is the first commercial-grade, open-source, distributed deep learning library written for Java and Scala. It allows users to compose flexible deep neural nets for production-grade frameworks, and was designed to bring AI to business environments for use on distributed GPUs and CPUs. Gibson created DL4J to make deep learning techniques available to anybody interested in them. It was also the driving force behind the launch of Skymind.
The Work of Skymind Global Ventures
Since cofounding Skymind and Deeplearning4j, Gibson has joined Skymind Global Ventures as vice president. At the heart of SGV is the idea of empowering AI ecosystems by helping companies to launch AI applications. This is done in three key ways: investing in proven startups, connecting them with opportunities by way of an in-house accelerator programme, and working closely with them to co-develop AI-driven products and solutions.
Gibson sees digital transformation as the major blocker for businesses and enterprises looking to implement AI technology. “Most (enterprises) still have many out of date IT practices,” he said, “making AI prohibitive to adopt.”
In terms of the work of SGV, which recently announced an $800 million fund to back AI companies and academic research, Gibson said, “SGV focuses on high-potential AI applications that can benefit from our software infrastructure expertise and partnerships in SE Asia. Initially, we are incubating projects internally. There is Konduit, and we have one other company that will be announced later.”
Konduit AI: Edge Devices for Business Applications
Konduit AI is focused on delivering Eclipse DL4J to clients and supporting their use of it through two key products: Konduit Edge and Konduit Appliance.
Konduit Edge is focused on deploying customized AI models onto edge devices, such as mobile or IoT. They offer a variety of models which are then customized for specific use-cases. The models run on Konduit Serving, and business metrics are monitored through a custom dashboard.
Gibson sees Konduit Edge as beneficial particularly for environments that are not friendly to large servers. “The top verticals are manufacturing and warehousing. In manufacturing you can use Edge AI to monitor assembly line output, as well as factory safety and factory equipment maintenance. In warehousing you can use it to predict demand based on what’s currently in stock and also use computer vision to analyze movement in warehouses. Detecting empty shelves/inventory for analysis purposes is also increasingly common to enhance automation.”
Konduit Appliance is an AI platform service for deploying AI models to servers. “It’s your more traditional on-premise AI model training and support,” said Gibson. “Basically, pre-configured hardware and software for running AI applications supported by Konduit. Hardware matters especially for AI applications; some customers just want to be able to run an intelligent application without knowing the difference between different kinds of CPUs, GPUs, and TPUs.”
As Konduit is an open-source company, their work can be easily found and accessed online. The links below are to some of their recent and current projects:
- Konduit Serving: A system and framework focused on deploying machine learning pipelines to production.
- Deeplearning4j: An open-source, distributed deep learning library for the JVM (Java Virtual Machine) that brings AI to business environments.
- JavaCPP: Software that provides efficient access to native C++ inside of Java.
- JavaCPP Presents: modules containing Java configuration and interface classes for widely used C/C++ libraries.
In a previous interview with Suzana Ilic from Machine Learning Tokyo, she mentioned the development of edge devices as a trend in Asia. Gibson also sees their potential. For him, “Edge will be widely deployed to reduce costs in running AI applications by running more compute locally. Edge devices are also typically cheaper than the combined cost of a centralized cloud plus a big server.”
AI Potential in Malaysia and Indonesia
As reported by TechCrunch in 2019, the Southeast Asia arm of Skymind signed a Memorandum of Understanding with Huawei Technologies to develop a Cloud and Artificial Intelligence Hub in Malaysia and Indonesia. Gibson sees great potential in these areas.
“Malaysia and Indonesia are two of the areas primed for cloud and AI growth in the future. The study from Microsoft and IDC Asia/Pacific indicates that by 2021, AI will double the rate of innovation in Malaysia and boost employee productivity by 60%. The study also found that 67% of business leaders and 64% of workers in Malaysia are positive about AI’s impact on the future of jobs. As for Indonesia, the market potential for AI is sizable. A large economy and population brings with it big challenges, but also many opportunities for AI solutions.
But it isn’t just Malaysia and Indonesia that show potential. “Outside of Malaysia and Indonesia, Singapore and Vietnam are primed for cloud and AI growth. According to Infocomm Media Development Authority, the AI market for Singapore has the potential to become a US$960 million market in 2022 and US$16 billion by 2030 with a compound annual growth rate of 42.2%.”
Recent Trends in AI Technology
When asked about trends in the development and application of AI technology, Gibson pointed towards recent work in healthcare technology. “Due to COVID-19 there is a lot of innovation in healthcare right now,” he said. “One such application is Axial AI, developed by our research lab partners in Shanghai.”
Axial AI is an assistive diagnosis tool, which utilizes AI technology for CT scan diagnosis. A joint development project led by the Shanghai Research Centre for Brain Science and Brain-Inspired Intelligence, Axial AI was trained on a dataset of 2000+ positive cases and 3000+ negative cases, and is used for triage, classification, and analysis.
To find out more about Konduit AI and their work in AI technology, you can visit their website here.
For more interviews with experts in the field of machine learning, be sure to check the related articles below. Subscribe to our newsletter for new interviews straight to your inbox.