ASVDA Calls on Startups to Show Innovative Energy at TIE to Help Speed Up Their Growth - EE Times

2022-11-14 14:44:28 By : Ms. Sue Su

Taiwan Innotech Expo (TIE) is an annual event for domestic industry players to demonstrate their technological strength. This year, showcasing their R&D capabilities in the “Innovation Pilot Pavilion” of Asia Silicon Valley Development Agency (ASVDA) are 13 Taiwanese high-tech start-up companies with innovative technologies such as artificial intelligence (AI), advanced displays, biochips, and privacy processing chips.

Integrating the resources of Taiwan’s top ten ministries and agencies, which include the Ministry of Economic Affairs, National Development Council, National Science Council, Council of Agriculture, Academia Sinica, etc., and gathering representatives from the industry, government, academia, and research, TIE aims to let the world know Taiwan, visit Taiwan, and invest in Taiwan. The ASVDA, on behalf of the National Development Council, invites startups and IoT-related vendors to take part in the exhibition every year. To find out and invite new rising starts to participate in the grand event, this year’s show will also follow the two major development axes of “Asia Silicon 2.0″, namely “to accelerate industrial evolution with AIoT” and “to drive next generation industry with innovation and entrepreneurship.”

Seeing the highlights of this year’s session, ASVDA Chief Administrative Officer Po-jung Lee pointed out that most of the startups were engaged in hardware manufacturing in the past; in contrast, many teams are now creating new services for data application and analysis. “Different from the past, we see more and more manufacturers using big data for back-end analysis and value-added applications. For example, about half of the 13 teams invited this time use AI technology to provide services. In recent years, we have also observed the rapid development of digital technology. It looks like the advances in data analysis technology have become a trend.”

Focusing on the market pulse, ASVDA announced last year (2021) it will upgrade ASVDP 1.0 to ASVDP 2.0, to further strengthen the R&D and application of AIoT technologies, improve the entrepreneurship ecosystem for startups and converge the output energy of the system as the three major strategies. Regarding the strategy to foster startup growth, ASVDA has already moved from “Start Up” to “Speed Up”, hoping not only to help startups gain a firm foothold but also to further integrate resources to accelerate their growth and to use systematic strategies to help them leap onto the international stage.

“Our expectations for startups are not the same as before,” said Po-jung Lee. As ASVDA has laid good foundations for market, talent, capital, regulations, and experimental fields over the past few years, there are now more than 200 new startup bases or terraces all over Taiwan, covering almost all the counties and cities. When resources reach a certain extent, it is more important now than in the past to focus on how to use the resources to improve the capacity and quality of services and adjust the existing resources to create greater synergy in the next stage. “We need to build a mechanism to integrate the efforts of those startups; they cannot fight alone as in the past. In the future, we hope to use the limited resources of the government to provide those startups with systematic assistance to reduce their market exploration time.”

The National Development Council has spared no effort in assisting startups to go overseas, including the creation of the national startup brand “Startup Island TAIWAN” and the promotion of the “Next Big” project, to form a startup national team and integrate resources to assist the team in expanding overseas markets. In addition, ASVDA also cooperates closely with Silicon Valley, the United States, and the Overseas Community Affairs Council (OCAC), hoping to gather resources from various parties to help Taiwanese manufacturers not only go beyond Taiwan but also expand in the international market.

Founded in 2019, Open AI Fab combines image vision and Natural Language Processing AI technologies to create a subscription-based medical AI business model. By using a large number of open-source AI algorithms, they customize AI models to assist customers in AI upgrades. At this exhibition, Open AI Fab will exhibit its technology for judging and identifying thoracic X-ray fractures.

As human life is sacred, the entry barriers in medical care are always very high. The reason why Open AI Fab can stand firm in the medical field is not only because Open AI Fab has recruited as its chief medical officer Wei-Yi Lee, an experienced emergency room doctor in the USA, but also because Open AI Fab masters the unique “Transfer Learning” technology as an important weapon to blaze the trail. Ming-ren Huang, CEO of Open AI Fab, made an analogy that a veteran Chinese cooking chef can usually earn other types of cooking faster; likewise, transfer learning enables AI models to own some basic recognition capabilities and use pre-trained models with a small amount of data to perform training and algorithm changes for different fields so that we can customize AI models with medical significance for teaching hospitals and mental health-related organizations.

On the basis of the existing AI, Open AI Fab has also been actively investing in the R&D of “Multi-Modal” technology in the past year. In the future, it will realize AI recognition of images, text, and voice. With the United States as its headquarters and Taiwan as its R&D base, Open AI Fab vows to continue expanding its business in Silicon Valley, USA under the name AI Fab Inc. The goal is to achieve a balance between profit and loss for its American company within two years. Chen-hua Yang, the co-founder, and head of business development of AI Fab Inc. revealed that he will begin by assisting the US Veterans Hospital and Dept of Veterans Affairs with closing in its Suicide Prevention goals by providing the suicide prevention coordinators with a 360-degree personality view of the patient through our AI-Empowered Matching Tool. Additionally, AI Fab will also provide its fractional AI Service to medical device manufacturers to help these companies become AI-enabled through customizable AI models in image, voice, and text processing.

What kind of expectations and imaginations can we have for the various displays that can be seen everywhere in our daily life? Taiwan’s Panel Semi, a startup only established in March 2021, uses its exclusive Active Matrix (AM) drive architecture and Mini LED direct display technology to create ultra-thin (1mm, which is about the thickness of a business card), lightweight, and low power consumption (no heat), eye-protecting flexible display. Not only can Panel Semi create a 110-inch super-large screen with a weight of only 20 kilograms by seamless splicing, but its flexible properties can also realize columnar screens and double-sided screens with various shapes, allowing the application of displays to break through the traditional framework and create infinite possibilities.

Although not yet 2 years old, Panel Semi’s team owns profound technical strength and rich industry experience, because most of its members were from InnoLux, a large Taiwanese display maker. Titus Chang, Vice President of Panel Semi, said that the company was born 10 years ago. At that time, the global LCD panel market was already caught in cut-throat competition. Thus, major Taiwanese panel suppliers decided not to “pivot” in the development of next-generation display technology while in order not to waste its previous efforts, the original team of Panel Semi set up its main direction based on the AM structure LED direct display technology. With this innovative mode, the team left InnoLux to join the TAcc+ (Taiwan Accelerator Plus) and finally achieved a small result after many tests.

Titus Chang revealed that in addition to displays, Panel Semi’s products and services also include AM drive architecture glass ICs, TFT antennas, and solutions for special panel designs and processes. Its glass ICs have been adopted by Japanese TV makers. Meanwhile, a number of overseas aviation companies are also optimistic about the thin, light, and energy-saving features of the Panel Semi display and are ready to apply it to public information displays in the cabin. As for its splicing large-size ultra-thin flexible displays and cylindrical displays exhibited at TIE this time, Panel Semi not only considers promoting it into the B2B market under its own brand but may also derive other innovative business models to open up a brand-new blue-ocean market, thus creating another sublime glory for Taiwanese panel suppliers in the global display market.

In the face of the persistent threat of the SARS-CoV-2 (COVID-19) variant virus, strengthening rapid detection capabilities has become one of the most effective means to prevent the spread of the pandemic. Although the commonly used household antigen rapid screening can obtain results in 15 minutes, it is prone to false positives/negatives; while the polymerase chain reaction (PCR) testing for COVID-19 is highly accurate, you have to wait at least 90 minutes for the results, which require the operation and interpretation of medical personnel. Therefore, the popularity is not high.

Molsentech has developed a highly sensitive “COVID-19 rapid detection chip”, which combines semiconductor and biomedical molecular detection technologies, suitable for the detection of viral nucleic acid or protein. The chip that can provide both the high accuracy of PCR testing and the quickness and simplicity of rapid screening helps to improve the early detection energy because the testing takes only 3 minutes at its fastest with an accuracy rate of over 95%.

Although many teams around the world have invested in the R&D of Bio-FET, it has not been commercialized yet. In contrast, Molsentech’s rapid detection chip was marketed after obtaining medical certification at the beginning of the year and continues to be supplied to medical institutions. Sheng-yuan Fan, Special Assistant to the CEO of Molsentech, pointed out that the biggest advantage in developing this technology platform in Taiwan, where the semiconductor industry is a niche, is that it can quickly get the response from the semiconductor factory and supply of components, helping to accelerate the R&D of this technology.

Sheng-yuan Fan also introduced that Molsentech adopts the technology related to nano-component biomedical testing that CEO Chia-jung Chu conducted research for 13 years at the Institute of Physics, Academia Sinica. Established in National Biotechnology Research Park in Zhubei in 2014, Molsentech used its core technologies of semiconductor wafer design and trace molecule detection to successfully develop a semiconductor biomedical wafer detection platform, which leads the world not only in the development of COVID-19 detection chips but also in the screening of diseases such as cancer, AIDS, and acute renal failure to implement early risk judgment and prevention.

The IoT market is a big cake for those who want to venture into it, but its fragmented, small-scale and diverse market properties makes most of the industry players difficult to enter IoT market smoothly. Daniel Lai, CEO of Brocere Electronics, said that the characteristics of IoT market make a high barrier for new industry players to make profit from IoT products. Therefore, to solve the pain point of entering the IoT market, Brocere has developed “Blockcraft,” an eco-system that only needs single device to meet all application demands of the IoT market.

Blockcraft modularizes the indispensable components in the IoT sensor architecture as building blocks, which helps customers to design end products flexibly to specified application. Daniel Lai explained that Blockcraft allows companies to eliminate the need to redevelop similar sensors for similar applications. In addition, because Blockcraft is composed by individual sensor and network modules, useless IoT Blockcraft sensors can be recycled and reassembled to use in other IoT applications. More importantly, Blockcraft can be powered by optional solar power module, thus contributing to the world’s efforts towards environmental sustainability as it helps conserve energy and reduce carbon emission.

It is worth noting that Brocere Electronics has formed Blockcraft Alliance since the Future Technology Exhibition, which Daniel Lai explained is meant to share the use of Blockcraft so that Taiwanese and global sensor module manufacturers can launch their sensor modules to enter IoT market faster and can link their data to the cloud with best quality. Blockcraft alliance can be a team to accelerate IoT products into various applications and scenarios. Aside from that, clients have chance to use more flexible IoT products and get a better user experience.

DeCloak’s De-Identification Technology Makes Private Data More Secure

In today’s era of Internet of Everything, whether it is to realize the identification function using physiological characteristics or unlock the device through a password, users may worry about whether their personal information is safely processed; for the industry players to manage such a huge amount of sensitive information, it also requires spending a great deal of effort and money to protect data well.

Fortunately, DeCloak Intelligences Co. uses de-identification technology to solve above-mentioned privacy concerns for both consumers and enterprises. DeCloak’s data privacy protection products cover both hardware and software – chips, AI algorithms, obfuscated image facial recognition, and searchable encryption engine. DeCloak’s co-founder and co-president Yao-Tung Tsou explained that the current face recognition technology is to take a photo of the user’s face and store it in the cloud or a centralized server database for identification, but DeCloak’s obfuscated image face recognition technology allows AI systems to authorize identities without storing sensitive face image data.

DeCloak’s co-founder and co-president Juang-Ying Chueh further explained that the obfuscated image face recognition is equipped with multi-factor authentication and adopts a zero-trust framework and a password-less method, so that users do not have to bear the data leakage risk when analyzing data.

More importantly, DeCloak’s AI model adopts data randomization and pixelization, which not only eliminates the possibility of data interception but also has a recognition accuracy rate of over 99%. Looking forward into the future, DeCloak will follow the pace of the industry to improve the data/image prediction accuracy of AI models and bring its new technologies to CES in 2023 to penetrate the huge data security and privacy market.

More of the information about TIE Exhibition

For more official review of startups’ information in detail, please refer ASVDA’s featured video:

(An advertisement of National Development Council)

You must Register or Login to post a comment.

This site uses Akismet to reduce spam. Learn how your comment data is processed.