From the Spring 2021 Edition of Discovery
USU Computer Science Alum Omar Flórez, PhD’12, Leads Efforts to Make Artificial Intelligence More Inclusive
This September, USU alum Omar U. Flórez serves as keynote speaker for the 2021 Tapia Conference, sponsored by the Association for Computing Machinery (ACM) and presented by the national Center for Minorities and People with Disabilities in Information Technology (CMDiT), the largest academic conference to celebrate diversity in computer science.
The virtual conference will reach computer scientists throughout the United States. It’s a special event for Flórez, who earned a doctorate from Utah State in 2012. Not only does it afford him with an opportunity to reach a national audience; it marks a full circle in his personal journey.
“The ACM Tapia Conference (named for Rice University mathematics professor and outreach activist Richard Tapia) supported me from the time I was a student,” Flórez says. “I have fond memories of attending the conference events and drawing inspiration from speakers from many countries and backgrounds. Now, I have an opportunity to give back.”
Flórez will speak about how language can be inclusive for everyone, as artificial intelligence (AI) can understand multiple languages simltaneously.
In the nine years since his USU graduation, Flórez, who is a natural language processing (NLP) research scientist at Twitter Cortex in San Francisco, has built an exciting and eclectic career. Not lost on him are the pathways and mentors who led him to enriching career opportunities in California’s information technology mecca from his hometown of Arequipa, Peru. Flórez is committed to opening those pathways to new generations of computer scientists.
A first-generation college student, Flórez received encouragement from his parents to excel in academics.
“I grew up in a middle class family and my parents never went to university,” he says. “When I wasyoung, they told me to not focus on money, but on knowledge.”
Peru, he says, has frequent earthquakes and, consequently, Peruvians have learned one’s assets can easily be lost to natural disasters. Rather than helping with the family business, Flórez’sparents encouraged him and his sister to focus on their studies.
“They just wanted us to learn as much as we could in school, hoping that would give us a different life,” Flórez says.
As a child, Flórez was fascinated by computers and robots. Following high school, he entered the Universidad Nacional de San Agustín de Arequipa, from which he earned a bachelor’s degree in engineering-computer science in 2006. During his undergrad years, Flórez wrote four research papers, accepted for presentation at refereed conferences.
“Doing research as an undergrad was fun and I learned a lot,” he says. “Later, I figured out that doing research was a good way to get into graduate school.”
Not finding the graduate programs he aspired to in Peru, Flórez looked abroad for opportunities.
“Peru is among the countries that consume technology, but it doesn’t produce a lot of it, so I soon realized I’d have to leave to pursue my passion,” he says. “One of the reasons I left my country, is because I wanted to do research in a global environment.”
He applied to graduate programs in Chile, Brazil and the United States, before selecting Utah State’s computer science program.
“Being willing to go somewhere else, beyond your comfort zone, to compete globally is very important for all students, no matter where you’re from,” Flórez says.
“I arrived at Utah State with very big dreams, because I wanted to be a really good professional in computer science,” Flórezsays. “It was a very different culture, but Cache Valley, with its friendly people, became my second home."
Flórez’s first USU advisor, Seungjin Lim was instrumental in recruiting the young student to Utah State.
“Dr. Lim left USU during my doctoral studies, but he was a very supportive advisor,” Flórez says. “He was the one who brought me to Utah State and I was amazed at how much he trusted in me, coming from so far away. I’m very grateful for his confidence in me.”
Flórez subsequently began research with advisor Curtis Dyreson, professor in USU’s Department of Computer Science.
“During my studies I learned mathematical techniques, algorithms and other computer science concepts,” he says. “But perhaps more importantly, Curtis taught me the importance of trust, allowing me the space and freedom to pursue ideas in research. This kind of level of confidence is important to achieve results.”
Dyreson correctly predicted future success for Flórez.
“He is very smart and independent,” said the professor during the scholar’s studies at Utah State. “He’s already made several contributions to the state of the art and I anticipate he’ll be a leader in this field in the future.”
While working with Dyreson, Flórez developed a novel research project employing the power of observation, probability and computation to give transit managers a leg up on heading off ugly traffic snarls. His efforts garnered a $20,000 IBM Scalable Data Analysis for a Smarter Planet Innovations Award, which propelled Flórez toward post-graduate opportunities.
The team’s proposal involved streamlining ways of searching and analyzing data from video traffic cameras to determine and predict patterns of commuter behavior. The aim of the project was to provide transit managers with reliable data to guide transportation management decisions, as well as access to real-time information to respond to problem situations.
While pursuing his degree at Utah State, Flórez, along with two friends from Peru attending graduate schools elsewhere in the U.S., developed the algorithm TalentScore,to seek out LatinAmerican youths through social media and match them with university scholarshipand assistantship opportunities.
“We knew how difficult it was to find, and gain access, to opportunities in higher education,particularly at the international level,” he says. “It’s even more difficult when you’re young and inexperienced, facing economic challenges, a first-generation student and, if you don’t have mentors.”
Flórez and his friends had already forged that path and wanted to share share what they’d learned. The trio called their creation “PrimerosPuestos”(First Places).
“The opportunities are there —it’s just finding a way to break through barriers that keep talented students and potential mentors from making a connection,” Flórez says. “But my friends and I believed we had just the tool to make those connections happen.”
In addition to helping teens and young adults find college-level opportunities, PrimerosPuestos connected primary schools in need with potential donors.
“We helped an elementary school in Peru collect donations to realize their dream of having a decent library,” Florezsays. “Our site allowed us to put schools in touch with appropriate donors who could quickly help. Our goal was to match each person or group in need with the best fit.”
The grad students’ innovation attracted attention, with their Facebook page ballooning to more than 12,000 fans within weeks. PrimerosPuestos was one of seven start-ups from Peru that advanced from the 2011 Intel Challenge Latin America competition to the international Intel Challenge. It was also a finalist in the 2011 Wayra competition,a contest sponsored by Spanish telecommunications company Telefónica to foster technological innovation throughout Latin America.
While the project received accolades from Telefónica and Intel, as well as welcome media exposure, it fell short of securing the needed funding to sustain itself and, within a year, folded.
“It was disappointing, but I learned a great deal from this venture,” Flórez says. “I was very inexperienced and social entrepreneurship was very new in Latin America. The investors we approached wanted to see a faster return on their investment than we could achieve. We had a longer-term view.”
He also realized the importance of timing.
“At the time, I thought, ‘We have a good team and a product based on good technology, so that should be enough,” Flórez says. “But I learned it takes more: You have to have the right ideas at the right time at the right place.”
From Utah State, Flórez traveled west, accepting a research scientist position with Intel Labs’ User Experience Research Team, based in Santa Clara, California. Among his initial assignments was working with the Anticipatory Computing Team to enable computers to understand causal relationships to explain human user behavior.
“For example, I implemented personalization algorithms from wearable devices that reused information, such as location, heart rate, physical activities, mode of transport, mood states, sleep quality and more, to treat people as individuals,” Flórez says. “I wanted to identify which patterns inpeople’s daily activities could negatively impact their sleep, so that a device could generate data-driven hypotheses that approximate a healthier lifestyle.”
From Intel, Flórez moved to Capital One, where he conducted research on enabling natural dialogues between humans and devices.
Currently at Twitter Cortex, Flórez is delving deeper into machine learning, studying how unlabeled collections of data contribute to understanding how people communicate in social media.
“We need lots of data to build these models,” he says. “On social media, people communicate via emojis, memes, entities, news and more. I study how to use all this information to understand and generate content, hoping computers will help to decode human language.”
Flórez is part of a committee assisting the Peruvian government in formulating a national artificial intelligence (AI) plan for Peru.
“Over the years, I’ve been able to contribute to my country, even though I live in the United States,” he says.
For context, Flórez explains, most countries are formulating AI plans.
“We see development in information technology as significant as the Industrial Revolution,” he says. “So, countries are analyzing their efforts and capabilities to align with goals for global competition, including investing in education, technology, communications and infrastructure.”
The idea, as well, Flórez says, is to be inclusive and to use technology to help everyone.
“For example, a homeless person should have access to the Internet, hardware and the opportunity to learn to code,” he says. “The government should make tools available for everyone to become educated and have the opportunity to get a job and achieve success in an information economy.”
The plan, Flórez says, is “connected to the idea of being able to compete globally, with no physical constraints –an idea I’ve embraced since my undergrad years.”
If artificial intelligence is to advance to the next level, the field needs many people and people of different backgrounds, Flórez says.
“I do research in machine learning and, because of that, I’m aware of the limits of artificial intelligence and where we may need to go in order to solve important problems,” he says. “We need many people to solve these challenges.”
The way to make this happen, Flórez says, is by having more diversity among computer scientists.
“I really encourage undergraduates to get into artificial intelligence and machine learning, because not only are there are a lot of opportunities to make progress in this area, but we need everyone to solve these challenges in fair and productive way,” he says. “We need to encode everyone’s view.”
Included in these efforts is recognizing not only barriers of culture, access and opportunity, but language. To that end, Flórez created a comprehensive Spanish-English dictionary of AI terminology. First shared in 2018 on the Internet hosting for software development site GitHub as an open source project, the dictionary, Diccionario Inglés-Español de Términos Técnicos en Inteligencia Artificial (IA)/Spanish-English Dictionary of Technical Terms in Artificial Intelligence (AI), allows users free access, along with the ability to add new terms to the growing project.
“The genesis of this dictionary came, in part, from my own experience,” he says. “Even though I speak English, it was challenging, when I was starting out in AI, to find technical documentation in Spanish and to map the new concepts in my second language.”
Flórez sees a Spanish-English dictionary as just a first step toward making the AI community more inclusive.
“I’d love to see similar dictionaries paired in other languages from Africa and Asia,” he says. “In a perfect world, everyone would have fair access to AI knowledge in their own languages, but the reality is English has become the lingua franca among AI developers. Multilingual AI dictionaries could help make AI information more readily available to the whole world, encouraging young engineers and students across the globe to get on a level playing field in AI development.”
Flórez says he’s greatly appreciative of his USU experiences and the mentors who guided him in his chosen path toward meaningful opportunities.
“I really want to say ‘thank you’ to all of my USU professors: Curtis Dyreson, Xiaojun Qi, Vladimir Kulyukin, Dan Watson and SeungJin Lim,” he says. “I have such good memories and we had such a good time together doing research.”
Flórez adds, “The confidence that had in me,when I was an inexperienced researcher, really left a mark with me. I’m very grateful.”
By Mary-Ann Muffoletto
Utah State University Professor and air quality expert Randy Martin was named Person of the Year at the Utah Clean Air Partnership Summit this past month, recognizing his work and achievements dedicated to improving Utah’s air quality.
The Utah State University rocketry team celebrated several wins at the annual Spaceport America Cup in White Sands, New Mexico.
The AeroLab at Utah State University is changing the way engineers design aeronautical vehicles with new, optimized design software built from scratch.
