From future wheat to speech recognition, HPC continues to enable innovations across a wide range of sectors
Facilitating and enhancing access of industrial actors to large-scale, high-performance computational (HPC) resources is one of the pillars of the EuroHPC Joint Undertaking to improve Europe’s competitiveness. HPC has been proven to drive significant returns on investments and enable innovation across various sectors. This article presents practical examples from some of the LUMI consortium countries on how HPC resources have been utilized in different research and development (R&D) projects in private as well as in public sector.
Belgium: Wheat of the future, state-of-the-art cooling components, intelligent document management
A team of data scientists at BASF use HPC & deep learning techniques to generate the wheat of the future. Their innovation center in Ghent, Belgium, aims to deliver solutions for BASF’s seed business by developing valuable traits for various crops through biotechnology solutions and modern germination techniques.
BASF looked into numerous ways to improve the germination process in testing new genetic lines to decrease the time-to-market for new wheat variations. Research activities at BASF have led to the creation of large amounts of wheat datasets.
Manually evaluating images from the field and greenhouse is a substantial task. It opens up a possibility to leverage on AI to accurately detect certain metrics (phenotypical markers) involved in the quality of the wheat plants using automated qualification. Recently, a team of data scientists at BASF started using deep learning for phenotypical quantification of biological images. As a result, they were able to add automation to the process, improve the final outcome with a higher confidence level and decrease the time-to-market for new wheat variations. Read more about the BASF success story.
Diabatix is a thermal design company that uses HPC and AI in its R&D development. The company designs cooling components such as heat sinks, liquid cooling channels, and cold plates using AI. The design process of the components is remarkably faster with the help of AI, and the features of these components are state-of-the-art. Have a look at the video:
A totally different HPC success story comes from Van Havermaet: a multidisciplinary company delivering, for example, accounting and legal services. This company takes advantage of HPC for its document management: it has around 15 million documents it needs to maintain and manage daily. Van Havermaet uses machine learning and AI to manage these documents, and with the help of HPC, it has managed to reduce the update cycle tremendously. Learn more about the case from the video:
Finland: More accurate speech recognition models paving the way for the era of voice control
Speechly, a Finnish start-up company established in Helsinki in 2016, has developed a new type of technology for understanding real-time spoken language. The patented machine learning technology combines speech-to-text with Natural Language Understanding in a novel way, enabling fast and more complex voice interactions than the current solutions on the market.
With the help of AI Computing Grant offered by Business Finland, the company got to use CSC’s Puhti computing environment for their product development. During the nine-month development project, Speechly improved the accuracy of its speech recognition model by almost 60% in terms of word error rate and was able to work faster and utilize larger sets of data than previously.
Speechly’s technology allows software developers to easily add an intuitive voice interface to any platform, whether it be Android, iOS, website, or video game. They have set the target high and believe that voice control will become more common in the same way as touch screens did after the iPhone was launched. Despite its size, this early-stage growth company is competing with the world’s largest tech giants, where some of Speechly’s people have worked in the past. The company is looking forward to work on LUMI, one of the world’s fastest supercomputers and most advanced platforms for AI that will soon start operating in Kajaani Finland.
Sweden: Sentiment analysis in Swedish
The National Library of Sweden (Kungliga Biblioteket, KB) has been awarded development access to the Vega EuroHPC JU system making it the first public administration actor in Sweden to access the system. The successful application was a joint effort between KB expert Dr. Love Börjeson and his group and ENCCS expert Dr. Hossein Ehteshami.
Sentiment analysis of texts and speech-to-text transformation are active areas of research and development in the field of AI. Two main ingredients of such endeavor are high-quality training data and a suitable deep neural network model, which uses the training data to tune its parameters. The reward is a system that not only can turn (almost) any speech to text but also “understand” the context and sentiment in it. Modern phones, laptops, and other gadgets are already using this technology to serve their owners. Nonetheless, most of the development in this field emerged around the English language model.
Currently, there is a void for a Swedish counterpart. As a response to this void, the data lab (KBLab) at the National Library of Sweden developed the KB-BERT model, the Swedish trained transformer model based on Google BERT architecture. KB-BERT, trained on the vast amount of high-quality data solely available at KB, proved to be a game-changer in this area.
With the awarded HPC time on the Vega EuroHPC JU petascale system within EuroHPC JU development call, training, and deploying of next generation of language models can be significantly facilitated.
Sweden: Better batteries for more sustainable future
Northvolt, a world leader in the field of battery production and development, has been awarded development access to the Vega petascale supercomputer hosted at The Institute of Information Science (IZUM) in Maribor, Slovenia.
This is a first in many ways: Vega is the first EuroHPC JU system to come online, and Northvolt will be the first European industrial actor to access a EuroHPC JU system. The nearly 2 million core-hours awarded will be used to perform larger and more realistic simulations of the electrochemistry relevant for battery development.
In partnership with ENCCS, the Swedish National Competence Center, Northvolt will investigate the use of classical and reactive molecular dynamics and quantum chemical simulations to devise bottom-up design strategies for improved batteries. This is an area of intense research with high societal impact: electric mobility has the potential to significantly reduce CO2 emissions of the road transportation sector, which currently accounts for about one-fifth of total CO2 emissions in Europe.
Up to one fifth of LUMI’s resources for industry
LUMI’s system deployment has progressed on time, and the first installation phase has been completed. The first phase of the system consists of the CPU partition, storage systems and an Early Access Platform for preparing applications for the GPU partition installed later this year. Up to one-fifth of the computing resources of the LUMI supercomputer will be available for companies. Learn more about what kind of benefits LUMI will bring to your company and stay tuned for information how to apply for LUMI’s resources!
Authors: Anu Märkälä, CSC, Tim Jaenen, FWO and Apostolos Vasileiadis, ENCCS
This is the part 2 of the article HPC continues to be a demonstrated enabler of innovation across a wide range of sectors, can drive significant returns on investments.