Sun Aug 30 2020
Microsoft Research Led Breakthroughs in Machine Learning
Machine learning is no longer taking the sidelines of technology and development. It has become an essential and indispensable part of businesses and consumers’ lives. Robert Schapire, a principal AI researcher for Microsoft, said in a post, "bit by bit, we’ve really been building this field from the bottom up, starting with basic problems. Machine learning has become applicable to such a huge array of problems. It’s really amazing."
Machine learning is the use of Artificial Intelligence (AI) to enable systems to learn and improve from experience automatically without requiring explicit programming. The goal is to create computer software that can gather data and learn from it to make precise, calculated decisions.
The learning process begins with observations from examples, direct experience, or pre-set instructions, to collect data and analyze patterns to make precise decisions in the future. The more data is collected, the better the decisions will be. It enables computers to learn automatically with little to no human intervention or assistance. It can augment what human analysts can do as they are provided with more precise tools. Human manpower can also delegate mundane, routinary tasks to machines and focus on other high-value tasks that do need much human intervention.
Microsoft is not just offering business, office, and home solutions. Microsoft has also taken leaps and bounds in innovations in technology like Artificial Intelligence and Machine Learning.
Microsoft engineers now strive to decentralize machine learning, so it is accessible for engineers and programmers who are still familiarizing themselves with machine learning. It allows for collaborative exchange between experts and non-experts, so the benefits of machine learning systems can be utilized widely.
The recent Microsoft-led Machine Learning breakthroughs are solving real-world problems and providing real-world solutions. Microsoft is aiming to bring these benefits to a broader array of people.
Let us look at some of these major breakthroughs in Machine Learning by Microsoft researchers:
Major Breakthroughs in Machine Learning led by Microsoft Researchers
1. Better AI translators
Thru machine learning, translating speech in real-time is now a reality. Microsoft aims to improve AI translators. In the annual Conference on Machine Translation (WMT) competition in 2019, Microsoft won in 8-19 challenges for language translation.
WMT competition, pushing through in November this year, endeavors to provide an avenue for academic researchers and tech companies globally to demonstrate their translation technologies in a competitive setting.
The Microsoft Research teams used innovative algorithms that improved the quality of results significantly in terms of machine-translation.
The implications of this are huge. Alexander Waibel of Carnegie Mellon University’s Language Technologies Institute mentions in a post that he is confident that machine-translation will reach human-level performance or even exceed, possibly within our lifetime. AI translation technology is, indeed, fast-evolving.
Other experts note, however, that translation technology is beneficial, but should not take the place of actual learning of languages. Machine learning-based translators still have a lot to improve on, but the use of AI and machine learning have remarkably increased the accuracy of the translation.
However, Linguists tell us that word meaning does not always match different languages. The context may be different, so what is said may not be translated with the same meaning or context. Nonverbal cues are also left out in translation. There are no reading-between-the-lines, which often means more than what is spoken. In machine translation, these judgments are not easily solved because, again, context varies across speakers and cultures.
Hand gestures are the simplest examples. Hand gestures are crucial for most people in a normal conversation. But hand gestures are non-verbal means of communication, and the meaning changes from country to country, culture to culture. Machine-translation can only cover the speech part of the entire communication process.
Still, this is a big leap in terms of machine translation. Technology is meant to help people, augment their capabilities, and make their lives easier. Machine-translation is the same. It is not intended to replace the totality of a communication process. But enhance it? Definitely.
There are so many applications where machine translation can bring many benefits, like in creative writing (grammar and readability accuracy), medical science, international relations (processing refugees, for instance),
The creation of a global, multinational, multicultural language is also possible with AI and machine learning. Imagine the language of the future, more universal in the sense that it will be a melting pot of common terminologies from different languages, and a mix of popularized or globalized terms coming from all across languages and dialects, on a bed of a primary language, like English, for instance. It may sound far-fetched now, but so did machine translation decades ago.
2. Better Facial Expression Recognition
Scientists at Microsoft are spearheading research on using AI and machine learning to build smarter apps able to recognize sounds, words, images, and facial expressions. Microsoft Project Oxford is going a step further with their research so that machines can identify even emotions.
Head of Microsoft Research Cambridge, Chris Bishop, demonstrated the emotion tool at the Microsoft Conference, Future Decoded.
Again, Microsoft is making its tools available to developers with no machine learning or artificial intelligence expertise. In this way, they can still incorporate capabilities such as speech, vision, and language understanding as they create their apps. Check this video out to grab a preview:
Microsoft Azure is one of Microsoft’s advanced touch-less solutions for businesses that offer advanced facial recognition according to diverse attributes. It is user-friendly, so developers with not much background in machine learning can easily add it to their apps with w single API call. It is also a cloud-based solution, so deployment is flexible. Programmers can run the program using cloud computing or edge computing. It also comes with enterprise-level security and privacy, so your data and trained models are protected.
3. Holo Lens 2
Walgreens is collaborating with Microsoft, using one of its latest product innovations, Microsoft HoloLens 2, in the training and development of their Kroger Express store employees in the US. Using Holo Lens 2, the employees are trained on how to properly restock products, conduct quality checks on perishable goods, and handle challenging customer service scenarios via a 3D mixed-reality model of their stores.
Microsoft partner Altoura intends to reacquaint employees with the store chain’s new layouts after renovations were done throughout the supply chain. The training is immersive and interactive, putting the employee immediately at the physical environment of the newly-renovated store, even before stepping in the actual location.
Walgreens will also collaborate with another Microsoft partner, Acuity, for an anonymous shopping path system. Bluetooth-powered carts and baskets gather data on how many customers are going through their remodeled stores chains. Microsoft then analyzes the data through Power BI to precisely measure if the new layouts are customer-friendly and efficient.
4. Health Technology
Have you ever thought about curing diseases, not with beakers and test tubes but computers and algorithms?
Technological health trends are increasing, and a group of Microsoft scientists, engineers, and programmers are endeavoring to develop computer science applications to aid in the fight against cancer.
Different teams are doing different strategies:
- Utilizing machine learning with natural language processing to assist oncologists in providing intuitive, accurate, and progressive cancer treatment efficiently for patients thru research data
- Coupling of machine learning and computer vision so radiologists have better scans and precise insights on progressing tumors in cancer patients
- The use of comprehensive algorithms, so scientists see how cancers develop and determine the best possible treatment to eradicate cancer in patients
- Giant leaps towards developing technologies that would allow scientists to program and control cells to combat diseases, including cancer
5. Continuity in Times of Crisis
In one of the most challenging pandemics in history, Microsoft aids in maintaining continuity and help in the world’s efforts to fight back and bounce back. Here are some of the ways that Microsoft is doing this:
- Covid-19 Health Bot: conversational AI tool allowing healthcare organizations to provide trusted and useful healthcare services and information. Immediate release of information is crucial in a time of crisis. More than 1,500 healthcare sites in 23 countries are using this tool. The Covid-19 Health Bot is helping more than 32 million people conduct a self-assessment of their symptoms.
- Computing power to health organizations: Microsoft is donating massive amounts of computing power to health organizations and research groups around the world. It is beneficial as the world works together to find the best possible solutions to combat the pandemic.
- Remote working: Microsoft solutions are supporting businesses and enterprises worldwide to facilitate remote working faster than expected. It is helpful for companies to be able to shift immediately to remote operations and distributed teams’ set-up to maintain continuity of service and operations.
Productivity and collaboration tools like Microsoft 365, running on the Microsoft Azure Cloud, leads the fleet of Microsoft’s solutions for touchless, cloud-based, and remote operations. Microsoft server and cloud-based storage solutions
Conclusion
Microsoft has brought the digital world many innovations through the years that are not just lofty and for a few. Its research keeps pushing boundaries of innovations while bringing these innovations towards the direction of the broader market as well. Indeed, it exhibits Corporate Social Responsibility (CSR) to a whole new level, that is why Microsoft is not only a household name but also a formidable influence in the world of technology today, including Artificial Intelligence and Machine Learning.
AUTHOR'S BIO
Mayleen Meñez
Mayleen Meñez worked for seven years in TV and Radio production, and also as a Graphic Artist/Editor. Finding her true passion, she devoted 15 years in NGO and community development work, where she experienced being a coordinator and teacher, travelling both in the Philippines and countries in Asia. She homeschools her three kids and reinvents Filipino dishes in her spare time. Writing has always been a hobby and pursuit, and she recently added content writing with Softvire Australia and Softvire New Zealand up her sleeve, while preparing for her next adventure in the nations.
