AsianScientist (Oct. 8, 2020) – First introduced in 1954 to meet the country’s need for skilled labor, polytechnics have played a pivotal role in Singapore’s successful industralization. But the world has profoundly changed since then, particularly following the invention of the integrated circuit in 1959. Today, powerful supercomputers with tens of thousands of chips are nothing out of the ordinary. With governments and industries alike increasingly facing complex problems answerable only with high performance computing (HPC), polytechnics must quickly adapt to equip their students with the skills to meet the latest industry needs.
To help their students stay ahead, Singapore Polytechnic (SP) has launched a full-time Diploma in Data Analytics and Artificial Intelligence (DAAA)—the first of its kind among the five polytechnics in Singapore.
But DAAA offered by the School of Computing (SoC) isn’t your run-of-the-mill data science/AI program. HPC, for example, features quite prominently in DAAA’s Practical AI module, enabling students to gain first-hand experience in a concept that had hitherto been thought inaccessible.
Building an HPC-ready workforce
“Many industries, especially consumer companies, are deploying machine learning and deep learning for their production lines,” explained Dr. Edna Chan, center director of SP’s Data Science and Analytics Centre, in an interview with Supercomputing Asia. “For such applications, especially those that are computationally expensive, they won’t be using normal servers and personal computers.”
Industry giants like General Electric (GE) and French Big Oil company Total have already begun investing in their own in-house supercomputing assets to improve their products and processes. GE, for instance, has used HPC to improve the design of their wind turbines, making them more aerodynamic and energy-efficient. Meanwhile, Total’s Pangea III—also known as the world’s largest industrial-use supercomputer—has been used to improve the detection accuracy of underground hydrocarbons.
Considering the close relationships between SP and its industry partners, Chan and her colleagues decided that it was high time for their students to gain the HPC skills desired by industry.
“It’s important for SP students to understand how HPC can be used in such situations and applications and be ready to make use of HPC when they enter the workforce,” she added.
This necessity for HPC skills was taken into consideration by SP as they developed the DAAA program. Hence, in the Practical AI module, DAAA students will specifically learn to perform the batch management and checkpointing of large-scale AI training jobs, allowing them to truly put HPC to the test. However, other students, like those from the Diploma in Computer Engineering (DCPE) will also get to join in the fun.
“As HPC hardware is typically housed in data centers, DCPE students will also learn to manage these centers,” said Chan.
High-powered help
Moving beyond industry, it’s worth noting that HPC is also no longer limited to advanced academic research laboratories. HPC has popped up in more mainstream applications, ranging from product design to driverless vehicles. Interestingly, some of these applications are currently being explored by SP’s various schools to address concerns on campus.
For example, a team from the School of Electrical & Electronic Engineering (SEEE) is working on a driverless electric vehicle that contains a bevy of sensors, including cameras, 5G and a light detection and ranging (LIDAR) system. Given the volume and complexity of the vehicle’s data, the team intends to use HPC to speed up computationally intensive processes like training the vehicle to recognize objects in its path. Once ready, the vehicle will be deployed around the SP campus to transport passengers and serve as an autonomous campus control platform. Given the current COVID-19 situation, the driverless electric vehicle could also double as a means to monitor and remind students, staff and visitors to adhere to precautionary measures like maintaining social distancing and wearing face masks.
SEEE has also developed a campus surveillance and building inspection system that utilizes 5G to analyze drone video footage in real-time. Just like the previous example, HPC will be used to train the complex images required by the system. With the help of HPC resources, Chan notes that SP can now take on more demanding projects and dream bigger, making the long-awaited move from mere experimentation to the level of minimum viable product.
Ahead of the pack
The School of Computing is planning to build two AI and Analytics Colabs equipped with 50 deep learning graphics processor units (GPU)-powered workstations and servers. Meanwhile, SEEE’s machine learning and AI laboratory will have 25 NVIDIA RTX GPU-powered workstations for training and modeling deep neural networks—an evolution of machine learning where brain-inspired artificial neural networks learn from large amounts of data. As indicated by its name, GPUs were originally intended to accelerate graphics tasks like rendering images, making them incredibly popular among avid video gamers. GPUs, however, have since found a second life in HPC due to their ability to carry out massive parallel processing.
While SP’s current assortment of equipment isn’t massive enough to be considered an HPC cluster, Chan shares that it should still allow students to appreciate the possibilities offered by HPC, especially when it comes to data-intensive modeling.
“Students can leverage HPC resources graciously provided by the National Supercomputing Centre when far more intensive computation is required,” she said.
While SP may be the oldest polytechnic in Singapore, its students are nonetheless kept up to date with a rapidly moving industry and even given a head start on HPC.
This article was first published in the print version of Supercomputing Asia, July 2020.
Click here to subscribe to Asian Scientist Magazine in print.
———
Copyright: Asian Scientist Magazine.
Disclaimer: This article does not necessarily reflect the views of AsianScientist or its staff.
