machine learning

Space doctor: U of T researcher co-chairs International Space University program in Brazil

mattimar — Wed, 26 Jul 2023 15:51:35 +0000

Space doctor: U of T researcher co-chairs International Space University program in Brazil

mattimar Wed, 07/26/2023 - 11:51

Cutline

From left: John Connolly, NASA lunar mission expert; Jonathan Clark, space medicine expert; Judith Hayes, NASA chief science officer; Safa Siddiqui, nuclear safety analysis engineer at Bruce Power; and Farhan Asrar, associate professor in U of T’s Temerty Faculty of Medicine (supplied photo)

Mariam Matti

Topic

Global Lens

Temerty Faculty of Medicine

Artificial Intelligence

Dalla Lana School of Public Health

machine learning

Medical Research

Space

Space exploration is often viewed as the realm of astronauts and engineers, but the ��Ƶ’s Farhan Asrar is using his expertise in public health to chart an extraterrestrial path in medicine.

Asrar, an associate professor in the department of family and community medicine in the Temerty Faculty of Medicine with a cross-appointment with the Dalla Lana School of Public Health, researches the physical and psychological effects of space travel on the human body, the challenges astronauts face during prolonged space missions and the potential implications for human health back here on Earth.

As a member of the International Space University, he recently travelled to Brazil to co-chair the department of human performance in space of the Space Studies Program, an annual nine-week program hosted in different locations around the world that provides participants professional development experience covering all aspects of space. As a co-chair, Asrar played a key role in developing curriculum for the program, organizing talks with astronauts and educational opportunities – including a trip to Brazil’s Paralympic training site to explore similarities between astronauts and elite athletes – and providing mentorship to attendees.

“I really appreciate that space brings everyone together – regardless of your background, culture or country,” the Mississauga-based physician and researcher says.

U of T News caught up with Asrar before he left for South America to ask him about his journey into space medicine, the challenges astronauts face and his role with the International Space University.

What is space medicine?

We are looking at what happens to astronauts and their health in space – we know that the microgravity has impacts on the body including bone and heart health. The space medicine side would introduce countermeasures that can maintain health and bone health in space, especially for someone who is staying for longer periods.

It also involves the preventive side because it is looking at individuals who might have medical conditions that would put them in a high-risk situation, and it examines whether their bodies would be able to withstand the strains of this extreme space environment.

What are some of the challenges astronauts encounter in space?

The distance and extreme environment can be stressful for anyone. When you are in space, you are away from your loved ones. You are in this really confined space with a select number of individuals for the next six months or so. Even simple tasks like going to the bathroom become more complex.

In order to prevent issues related to bone and heart health, astronauts have to exercise for two hours on a daily basis. They also have their regular schedule like keeping their premises clean and research projects – and working to meet all those tasks and needs in an extreme environment can be quite challenging.

With further distances away from Earth, the time delay creates challenges in communication. If someone has a medical emergency or needs mental health support, it can be difficult to communicate in a timely manner. These are the sorts of things we’re examining in space medicine: how do we train people to be better prepared for these challenges?

As a member of the Temerty Centre for AI Research and Education in Medicine (T-CAIREM), I know that there are developments in machine learning and artificial intelligence that are being explored for use in every aspect of medicine. This also has a role in space medicine and improving the response time from Earth to astronauts.

Farhan Asrar leads an introductory lecture on space medicine for Space Studies Program participants (supplied photo)

What sort of work will you be doing with the International Space University’s Space Studies Program?

This year, the space studies program is being held in Brazil and I was invited to co-chair the department of human performance and space along with Judith Hayes, NASA's chief science officer under the human health and performance directorate.

We helped develop curriculum for the program along with other chairs, directors and academics. We’ve organized talks, educational activities and different sessions that are related to human performance and health and medicine in space. We’ll be there to support and mentor the participants when necessary.

We’ve invited CSA’s flight surgeon and NASA’s Space psychologist to share their expertise on the health of astronauts, the training they do to prepare for missions and the challenges they face. We’ve also invited NASA astronaut Jessica Meir and Canadian astronaut David Saint-Jacques to talk about his experience in space. He has worked as a family doctor in the Arctic, so it’ll be interesting to hear about his experiences with extreme environments and challenges he’s had here on Earth, too.

I’ve also been researching the link between sports and space medicine – so we organized a visit to Brazil’s Paralympic training site. There are a lot of similarities between athletes and astronauts: they both have to be physically fit, train every day, and their training and professions put a physical strain on their bodies. In this session, we’ll also educate participants about Paralympic training and then get into physical barriers in space.

What excites you most about space?

I really appreciate that space brings everyone together, regardless of your background, culture or country. It’s great to have so many individuals working together – even the crew missions are now more diverse.

There’s also a lot of technology like GPS and infrared thermometers that originated from space programs before they became part of our everyday lives. Infrared thermometers, for example, were invented to measure the temperature of distant celestial objects and become very popular during the pandemic to check a person’s temperature.

Similarly, the isolation and extreme environments that astronauts feel on their missions is something we felt during COVID-19. And we used telemedicine during the pandemic, which originated from space and taking care of astronauts at a distance. With the pandemic, we were put in many situations we weren’t used to – but you could often look to the space world and say, “Yes, we’ve been doing this for decades and we can share our expertise.”

There’s so much that space introduced to our lives that we don’t even realize.

Researchers accelerate development of cellular therapies for damaged tissues

siddiq22 — Tue, 25 Jul 2023 16:17:02 +0000

Researchers accelerate development of cellular therapies for damaged tissues

siddiq22 Tue, 07/25/2023 - 12:17

Cutline

From left: PhD candidate Oreoluwa Kolade, Professor Julie Audet and Assistant Professor Sowmya Viswanathan are working to identify the optimal cell culture conditions for different donors without doing full sets of detailed experiments (photo by Safa Jinje)

Safa Jinje

Topic

Breaking Research

Data Sciences Institute

Institute of Biomedical Engineering

Temerty Faculty of Medicine

Faculty of Applied Science & Engineering

Graduate Students

machine learning

Regenerative Medicine

Research & Innovation

Subheadline

Optimizing conditions to culture therapeutic cells can help reduce the costs and labour of experiments

Innovations in the ways that human cells are grown in laboratories could help speed up the development of cellular therapy, a branch of regenerative medicine that targets diseases that are currently incurable.

Julie Audet, a professor in the ��Ƶ's Institute of Biomedical Engineering in the Faculty of Applied Science & Engineering, is working to address some of the most significant challenges related to producing therapeutic cells.

Her research on cell and tissue engineering aims to enhance the therapeutic properties of lab-grown human cells to ensure that they are ready for clinical application.

Audet and her team are developing complex computational algorithms to optimize laboratory experiments for academic and industrial researchers. These tools will allow researchers to create the best conditions to culture therapeutic cells.

Cellular therapy acts by transplanting enhanced human cells into the body to replace or repair damaged tissue and cells, in order to treat a variety of diseases and conditions – an approach sometimes referred to as a living drug.

“But before cells can be transplanted into a patient, we want to ensure that the cells are not contaminated with compounds that can trigger an adverse reaction,” Audet says.

“We also need to enhance the therapeutic properties of the cells in a culture process to effect a positive medical outcome.”

There are many factors to consider when designing cell-culture experiments. For example, there are numerous expensive reagents to select and optimize at different doses – these substances are used to test chemical reactions carried out by the cell. There are also significant technical and biological variations to consider – cells from different donors don’t always behave the same way in culture.

“Our algorithms become necessary when experiments are very costly to execute and are extremely labour-intensive,” Audet says.

“They are especially useful when the results are impossible to predict because of the complexity of the biological systems under study. In that case, it would not be feasible or possible for researchers to use conventional approaches to design and execute their experiments.”

With complex computational algorithms based on machine learning, Audet and her team can design experiments that are not only feasible and offer a greater chance of success, but are also less costly, with fewer resources needed to execute the experiment.

An earlier prototype of such algorithms was used by Audet’s lab to make a serum-free T-cell medium to treat blood disorders. The algorithm was also used by Craig Simmons, a professor in the Institute of Biomedical Engineering and the department of mechanical and industrial engineering in the Faculty of Applied Science & Engineering, and Neal Callaghan, a PhD graduate from the institute and a former researcher in Simmons' lab, to develop culture media for cardiomyocytes (cardiac muscle cells), a process that is now being commercialized.

“Cardiac tissue engineering is an important application for our tools because when it comes to heart failure and heart disease, there are many conditions that can’t currently be cured,” Audet says.

“Cellular therapy offers a promising approach to treat heart failure and other cardiac ailments.” 

Audet is working with Sowmya Viswanathan, a researcher at the University Health Network's Krembil Research Institute and an assistant professor in the Institute of Biomedical Engineering and the Temerty Faculty of Medicine, on research that addresses the distinct characteristics of cell donors that make cells behave differently in culture.

Viswanathan is developing cellular therapies using mesenchymal stromal cells to combat osteoarthritis, a chronic inflammatory disease.

“We have seen that different mesenchymal stromal cell donors prefer different growth conditions,” Viswanathan says. “This algorithm helps us identify optimal conditions for different donors without doing a full set of detailed experiments.”

The goal of her collaboration with Audet is to also develop categories of conditions that could be matched to diverse groups of donors based on genetic markers. 

Audet and Viswanathan co-supervise Oreoluwa Kolade, a PhD candidate at the Institute of Biomedical Engineering and a fellow at the Data Sciences Institute – one of several U of T institutional strategic initiatives – whose work brings together tissue engineering and data science.

“My research involves the statistical design of experiments, looking at the numerous factors that can impact the cell expansion process, such as oxygen levels, cell density and medium composition,” Kolade says.

“If a researcher wants to see which factors maximize the therapeutic quality of their cells, it isn’t cost-effective to test all possible combinations. So we are trying to design experiments in such a way that researchers can get the highest impact when they see their results.”

This model would allow researchers to narrow down the varying factors and run combinations into a simulation model to see which experiments they would need to do to get the best cells.

“We are currently working on the commercialization of the latest version of one of the algorithms that includes these machine-learning modules to help design experiments,” Audet says.

“We hope this tool will help make cellular therapies both widely available and more accessible by accelerating the development of these therapies while increasing the effectiveness of the enhanced cells.”

New autonomous lab at ��Ƶ to improve drug formulation

siddiq22 — Fri, 14 Jul 2023 20:09:01 +0000

New autonomous lab at ��Ƶ to improve drug formulation

siddiq22 Fri, 07/14/2023 - 16:09

Cutline

From left: PhD trainee Zeqing Bao, Professor Christine Allen, Allen Lab Director of Research and Partnerships Pauric Bannigan (photo by Steve Southon)

Eileen Hoftyzer

Topic

Breaking Research

Acceleration Consortium

drug development

Institutional Strategic Initiatives

self-driving labs

Artificial Intelligence

Entrepreneneurship

Leslie Dan Faculty of Pharmacy

machine learning

Research & Innovation

Subheadline

Facility at Leslie Dan Faculty of Pharmacy is one of six at U of T being funded by $200-million grant to Acceleration Consortium

A new autonomous lab being built in the ��Ƶ's Leslie Dan Faculty of Pharmacy will help to design and optimize formulations that will improve bioavailability, stability and efficacy of a variety of drugs.

Christine Allen, a professor in the faculty whose research focuses on drug development and disease diagnostics, is co-leading the lab with Frank Gu, a professor in the department of chemical engineering and applied chemistry in the Faculty of Applied Science & Engineering.

“I see huge potential for artificial intelligence (AI), machine learning and automation in pharmaceutical sciences, for formulations and beyond,” says Allen, an expert in drug formulations.

“The world is finally understanding the impact of formulation technology and how powerful it is. Now we can marry that technology with AI and machine learning, so we’re kind of unstoppable.”

The new facility is known as a 'self-driving' lab because it uses AI, automation and advanced computing to test different combinations of materials and iteratively develop the best formulations. This approach will allow researchers to evaluate a much larger number of materials, significantly reducing both the cost and time required to identify a new drug formulation.

Allen explains that formulation scientists are often working under tight time constraints to get a product to market by deadline, and the final formulations are the best that can be done with the time available – but are not necessarily the best possible option.

“We’re providing a solution to that – accelerating development while looking for the best formulation,” says Allen, whose team worked in collaboration with Professor Alán Aspuru-Guzik’s research group to develop a prototype of the self-driving lab last year.

The self-driving lab at the Leslie Dan Faculty of Pharmacy is one of six being built at U of T through a $200-million Canada First Research Excellence Fund grant to the Acceleration Consortium, a global network of government, industry and academic researchers accelerating the discovery of materials and molecules needed for a sustainable future.

Researchers say the self-driving lab will help the pharmaceutical industry improve efficiency in drug development and production (photo by Steve Southon)

Allen played a key role in launching the Acceleration Consortium during her time as associate vice-president and vice-provost, strategic initiatives.

As a co-lead on the project, Gu is bringing his expertise in nanotechnology engineering to develop high-quality, precise formulations that use nanotechnology to improve the delivery, bioavailability and efficacy of drugs.

“Nanotechnology engineering has revolutionized the pharmaceutical industry, and its potential applications in a self-driving lab are both imminent and paradigm-shifting,” Gu says.

“By harnessing the power of nanotechnology in a self-driving lab for pharmaceutical formulations, my lab is working with Professor Allen’s team to unlock many advantages that will transform drug development, formulation, modular manufacturing and patient care.”

Allen says that the lab's strong collaboration between pharmaceutical scientists and computer scientists also provides unique educational opportunities for trainees that will provide them with key skills needed for a career in the pharmaceutical industry.

“The trainees are experts in both fields – this is where the future is going, and that to me is the power in all of this. It is true interdisciplinary collaboration,” Allen says. “The trainees are gaining knowledge, expertise and experience, and they are then able to secure exciting positions in industry and be leaders in the field.”

Last year, Allen took a leave of absence from the university to take on a leadership role with adMare Bioinnovations, an organization that helps support Canadian life science companies and researchers. She has now returned to the Leslie Dan Faculty of Pharmacy full-time and brings back a wealth of new experience in entrepreneurship that she plans to apply to her research and the work of the Acceleration Consortium.

To that end, Allen is also chairing the Acceleration Consortium's committee on commercialization and partnerships, and is also developing a graduate course in innovation and entrepreneurship.

With the lab's focus on translational research and commercialization, both Allen and Gu are excited about its potential to generate new formulations that will ultimately improve patient outcomes.

“The self-driving lab is the most cutting-edge technology to help the pharmaceutical industry to drive toward digitization, automation and ultimately improve efficiency in product development and production,” Gu says.

“With its unique capability in bridging automation and iterative discovery process, it is also the key to helping us accelerate the development and production of personalized medicines, which is prohibitively expensive and labour-intensive to be done today.”

Researchers explore shifting AI landscape at Absolutely Interdisciplinary conference

siddiq22 — Wed, 12 Jul 2023 19:00:25 +0000

Researchers explore shifting AI landscape at Absolutely Interdisciplinary conference

siddiq22 Wed, 07/12/2023 - 15:00

Cutline

Participants at the Schwartz Reisman Institute for Technology and Society’s annual conference, Absolutely Interdisciplinary, discussed what AI can teach us about social systems, cognition, education, creativity and more (supplied image)

Topic

Our Community

Schwartz Reisman Institute for Technology and Society

Artificial Intelligence

Centre for Ethics

Computer Science

machine learning

Ontario Institute for Studies in Education

Psychology

Research & Innovation

Researchers across disciplines gathered recently at the Absolutely Interdisciplinary conference to reckon with the ramifications of the evolving AI landscape – from the technology’s transformative impacts on work and education, to its implications for human cognition and values.

The third annual academic conference held by the Schwartz Reisman Institute for Technology and Society (SRI) at the ��Ƶ featured 23 speakers from diverse fields such as computer science, psychology, law, economics, education, philosophy, media studies, and literature.

“The recent impacts of generative AI tools have really emphasized how important it is to build spaces for interdisciplinary conversations,” said SRI Director and Chair Gillian Hadfield, the inaugural Schwartz Reisman Chair in Technology and Society, CIFAR AI Chair and professor in the Faculty of Law and Rotman School of Management.

“Our goal with Absolutely Interdisciplinary is to foster new research agendas based in the creative interplay of diverse questions and framings as we explore the potentials of these new technologies.”

Cognition, storytelling and the future of intelligence

The conference featured an opening keynote by Blaise Agüera y Arcas, vice-president and fellow at Google Research, and a regular participant in developing cross-disciplinary dialogues about AI and ethics, fairness and bias, policy and risk.

“We’re all going through a bit of existential angst because most of us believe that we have been the smartest things on Earth for some time now,” he said. “I don’t think that will any longer be the case.”

In his keynote lecture, Blaise Agüera y Arcas of Google Research spoke about the history and potential future of AI development (supplied image)

Exploring the historical connections between neuroscience, computer science and cognitive science, Agüera y Arcas discussed the progress made in visual perception by neural networks in recent years, and delved into the concept of artificial general intelligence (AGI), including ongoing debates on whether AI might become sentient and pose a risk to humans.

“Some people take AGI to mean superintelligence, others take it to mean consciousness,” said Agüera y Arcas, who highlighted storytelling as an essential quality of human experience, and how the construction of personal narratives about ourselves and our past help to predict our future.

“We are the stories that we tell ourselves,” Agüera y Arca noted. “And as we interact with people, we construct and edit that story over time.”

AI, learning and the role of education

In a session pairing psychology with computational neuroscience, Joel Leibo of Google DeepMind and SRI Faculty Fellow William Cunningham, a professor in U of T’s department of psychology, presented their ongoing work on modeling human social interactions using artificial agents to test social cognitive theory.

Leibo and Cunningham showed that by simulating coordination games in multi-agent reinforcement learning, they could test the origins of very human things like in-group bias and coalition-building.

SRI Research Lead Ashton Anderson, assistant professor in the department of computer science, moderated another session on learning from a different perspective: the implications of generative AI tools such as ChatGPT for educators and students.

Panelist Lauren Bialystok, an associate professor of social justice education at the Ontario Institute for Studies in Education and acting director of U of T’s Centre for Ethics, offered ideas on why we value the concept of “originality” and discipline students who violate it by using ChatGPT to write assignments.

“What is the benchmark against which cheating emerges as a moral wrong or a pedagogical error?” asked Bialystok.

“Is technology the enemy of originality? What about individual originality versus collective originality? We need to start sussing out what really matters to us in student learning and student assessment.”

From left to right: Moderator Nicolas Papernot with panelists William Cunningham and Joel Leibo, who described how multi-agent reinforcement learning can be used to test aspects of social cognitive theory (supplied image)

AI’s capabilities, behaviours and harms

One of the more frequent fears cited about AI is that it will replace workers – a subject tackled in a session moderated by SRI Research Lead Avi Goldfarb, the Rotman Chair in Artificial Intelligence and Healthcare and a professor of marketing at the Rotman School of Management. Economist Daniel Rock of the University of Pennsylvania’s Wharton School shared his recent findings on which occupations are most likely to be impacted by large language models.

“People who process information [and] have knowledge as part of their work are more exposed,” said Rock, who noted that this exposure can be harmful or helpful.

“One of the key things for economists to add to this conversation is equilibrium,” Rock said. “It’s not just about AI replacing workers – there’s complementary innovation to be done here; there’s supply and demand; there’s the question of whether making one part of work cheap makes another part of it very expensive. There is a lot more work to be done here.”

New languages, new frameworks for understanding

The conference closed with a session on AI and creativity led by SRI Faculty Fellow Avery Slater, an assistant professor in the department of English in the Faculty of Arts & Science. The session featured presentations by literary scholar N. Katherine Hayles, whose work has focused on relations between science, literature and technology, and U.K.-based poet Polly Denny, who presented her experiments with text-generating AI systems that have yielded new forms of artistic collaboration.

Where Agüera y Arcas’s opening keynote highlighted the role of narration about the self as constitutive of what we might think of as AGI, Hayles also framed the question of whether AI is “alive” in similarly fundamental terms of how perception and meaning-making construct and define the environments in which we operate.

“Where there's life, there’s cognition. But computational media also have cognitive capabilities,” Hayles said.

Read the full story at the Schwartz Reisman Institute

How AI and neuromodulation could help with sleep disorders

siddiq22 — Tue, 11 Jul 2023 18:37:12 +0000

How AI and neuromodulation could help with sleep disorders

siddiq22 Tue, 07/11/2023 - 14:37

Cutline

Assistant Professor Xilin Liu and his collaborators are developing electronic devices that could help patients suffering from sleep disorders (photo courtesy Xilin Liu)

Selah Katona

Topic

Breaking Research

Artificial Intelligence

Faculty of Applied Science & Engineering

machine learning

Neurology

Research & Innovation

Sleep

Subheadline

'We still don’t fully comprehend what actually occurs in our brains during sleep'

A new partnership between the ��Ƶ's Xilin Liu, assistant professor in the Edward S. Rogers Sr. department of electrical and computer engineering in the Faculty of Applied Science and Engineering, and Andrew G. Richardson, research assistant professor of neurosurgery at the University of Pennsylvania, will develop a new generation of electronic devices to investigate sleep modulation.

Their research will potentially develop new interventions that help deal with a wide range of sleep disorders.

On average, we spend a third of our life asleep. During sleep, the brain undergoes important processes that support memory consolidation, neural restoration and the clearance of toxins. Sleep disruptions can interfere with these processes. But while good "sleep hygiene" is increasingly recognized as crucial to both physical and mental health, sleep disorders remain widespread.

“40 per cent of Canadians have sleep disorders, with over 3 million suffering from insomnia,” says Liu, director of the X-Lab and affiliate scientist at the the KITE Research Institute.

“Sleep deficits negatively affect brain functions such as attention and memory, and immune function, metabolism and heart health. Chronic sleep-wake disruptions are connected to neurodegenerative disorders such as Huntington’s, Parkinson’s and Alzheimer’s disease, and cognitive decline with aging.”

This diagram describes the process of closed-loop sleep modulation via miniaturized electronics
(image courtesy of Xilin Liu)

Liu’s research focuses on developing integrated circuits and systems for advancing health care, digital communication and machine learning. In the new collaboration, he will be building fully integrated wireless systems-on-chips that can autonomously modulate sleep behavior in pre-clinical studies.

“Sleep is a complex procedure involving different stages and patterns,” he says. “To address this, we are integrating machine-learning algorithms into our devices. These new algorithms can recognize sleep patterns and identify sleep disorders that may not be distinguishable using traditional algorithms.”

Liu and his collaborators hope that these new approaches will enable them to gain a deeper understanding of how our brains function during sleep and how to modulate sleep circuits.

They also plan to incorporate various neural interfacing capabilities into the system, which will enable more accurate and precise interventions. The results of this research will contribute to the advancement of neuromodulation, a technology that involves placing devices inside a patient’s brain, spinal cord or peripheral nerves. These devices are designed to regulate neural activity and help reduce symptoms related to different disorders.

Liu is a faculty member of the CRANIA Neuromodulation Institute (CNMI) in the Faculty of Applied Science and Engineering, which brings together experts in engineering and neuroscience in a collaborative hub for neuromodulation research.

Liu’s project was recently awarded $2.2 million by the National Institutes of Health (NIH) through their Research Project Grant Program (R01), and is supported by industry partners such as the Canadian Microelectronics Corporation and Open Ephys.

The NIH R01 grant is a highly competitive award and Liu and Richardson’s proposal fell in the top 1 per cent of applications.

“This is a great sign that NIH recognizes the value and impact of this research and the caliber of the team,” Liu says. “We’re excited to receive this funding and over the next four years we hope to get to a stage where the technology can be used in clinical trials.”

Liu’s research team has a long-term goal of creating wearable devices for sleep modulation that people can use at home to enhance the quality of their sleep.

“While there are medications available to treat sleep disorders, the challenge lies in the fact that we still don’t fully comprehend what actually occurs in our brains during sleep,” Liu says.

“It’s possible that there are low-cost, high-efficacy treatments available for sleep disorders that we are currently unaware of.”

From machine learning to mentorship, graduate Irene Fang showed leadership during her time at U of T

siddiq22 — Thu, 22 Jun 2023 17:43:55 +0000

From machine learning to mentorship, graduate Irene Fang showed leadership during her time at U of T

siddiq22 Thu, 06/22/2023 - 13:43

Cutline

Irene Fang graduated with an honours bachelor of science degree, working on research that could lead to new treatments and therapies for immunocompromised patients (supplied photo)

Topic

Artificial Intelligence

Faculty of Arts & Science

Immunology

machine learning

Medical Research

Subheadline

Even while undertaking complex research, the human biology and immunology student took the time to help her peers

While studying for her honours bachelor of science degree, new ��Ƶ graduate Irene Fang capitalized on opportunities both inside and outside the classroom.

Majoring in human biology and immunology in the Faculty of Arts & Science, Fang researched innovative methods in ultrasound detection driven by artificial intelligence (AI) and machine learning. She’s also working on research into cells and proteins in humans that could lead to new treatments and therapies for immunocompromised patients.

Even amid that busy schedule, Fang was determined to help others succeed. As a senior academic peer advisor with Trinity College, she was admired for her dedication to learning and the U of T community.

“I want to keep giving back because I am so appreciative of the upper-year mentors I connected with, starting in first year,” Fang says. “They continue to serve as an inspiration, motivating me to further develop personal and professional skills.”

Fang spoke with Faculty of Arts & Science writer David Goldberg about what she learned during her undergraduate studies, the importance of peer support and her post-graduation plans.

Why was U of T the right place for you to earn your undergraduate degree?

U of T provided a plethora of academic, research and experiential learning opportunities alongside a world-class faculty to help cultivate my curiosity and consolidate my knowledge. In conjunction with an unparalleled classroom experience, I gained a real-world perspective with international considerations through the Research Opportunities Program.

I would be remiss if I didn’t also mention how extracurricular activities enhanced and enriched my university experience. The many clubs at U of T helped me focus on my passions and make meaningful connections with like-minded peers who became my support network, enabling me to reach my full potential.

How is your area of study going to improve the life of the average person?

It is absolutely fascinating that AI has already revolutionized the medical field. Specifically, AI possesses the potential to aid in the classification of ultrasound images, enhancing early detection and diagnosis of internal bleeding because of injuries or hemophilia. Overall, AI may lead to more efficient care for patients, thereby improving health outcomes.

In terms of my immunology research, since the memory B cells expressing the specific receptor are dysregulated in people suffering from some autoimmune disorders and infectious diseases, a better understanding of how memory B cells are regulated could provide valuable insight into the underlying mechanisms of such diseases so we can enable scientists to develop new therapies that alleviate patients’ symptoms.

What are you hoping to do after graduation?

I aspire to pursue a career in the medical field, conduct more research and nurture my profound enthusiasm for science while interacting with a diverse group of people. I hope to devote my career to improving human health outcomes while engaging in knowledge translation to make science more accessible to everyone.

Why was working as a peer advisor at U of T important to you?

I remember feeling overwhelmed as a first-year student until I reached out to my academic peer advisors. Had I not chatted with them, I would not have known about – let alone applied for – my first research program. Looking back, it opened the door to many more new, incredible possibilities and opportunities.

This experience made me realize the significance and power of mentorship, inspiring me to become an academic peer advisor. Seeing my mentees thrive and achieve their goals has made this role so rewarding – so much so that I am determined to engage in mentorship throughout my career after graduation.

What advice do you have for current and incoming students to get the most out of their U of T experience?

Ask all questions – because there are no silly questions. Get involved, whether it be volunteering, partaking in work-study programs, sports or joining a club. Meeting new people and talking to strangers can be daunting, but the undergraduate career is a journey of exploration, learning and growth.

Be open-minded and don’t be afraid to try something new. Immersing yourself in distinct fields enables you to discover your interests and passions, which can lead you to an unexpected but meaningful path.

Also, be kind to yourself because failures are a normal part of the learning process – what’s important is that you take it as an opportunity to learn, grow and bolster your resilience.

And finally, although academia and work can keep you busy, remember to allocate time for self-care. Exercise, sleep and pursue hobbies because mental health is integral for success in life.

New PhD graduate Guodong Zhang hopes to lead the next generation of AI experts

Christopher.Sorensen — Wed, 07 Jun 2023 16:05:23 +0000

New PhD graduate Guodong Zhang hopes to lead the next generation of AI experts

Christopher.Sorensen Wed, 06/07/2023 - 12:05

Cutline

Graduating PhD student Guodong Zhang's work on theoretical foundations and practical algorithms for machine learning have already been adopted by major players in the AI sector (supplied image)

David Goldberg

Topic

Our Community

Convocation 2023

Artificial Intelligence

Computer Science

Faculty of Arts & Science

Subheadline

Zhang decided to pursue graduate studies in computer science at U of T after being inspired by AI pioneer ��Ƶ

As issues around the usage and ethics of artificial intelligence (AI) continue to evolve, new ��Ƶ graduate Guodong Zhang is ready to address those challenges.

“The public should embrace AI as a catalyst for positive change,” says Zhang, who graduates this month with his PhD from the department of computer science in the Faculty of Arts & Science.

“Ensuring that AI systems align with human values and remain under human control becomes increasingly critical. Addressing AI safety is one of the most important and impactful problems we face today.”

Zhang intends to be part of the solution. He applied to U of T for his PhD several years ago, inspired by the “godfather of AI” ��Ƶ, University Professor emeritus in the department.

As a PhD student, Zhang taught many courses at U of T. His study of theoretical foundations and practical algorithms for machine learning have already been adopted by major AI players, including Google Brain, DeepMind and OpenAI.

Ahead of convocation, Faculty of Arts & Science writer David Goldberg spoke with Zhang about his research and how U of T prepared him for a future career in AI.

Why was U of T the best place to earn your PhD?

I was captivated by the immense potential of deep learning, so U of T was an obvious choice for me given its leadership in this area. ��Ƶ and his students shocked the world with their results on ImageNet in 2012 with AlexNets, a neural network architecture which started a golden age for deep learning. Furthermore, the prospect of collaborating with the Vector Institute rendered U of T even more special.

How do you explain your work with AI to people outside your field?

I focused on developing neural network models and algorithms that excel in fast training, robust generalization and accurate uncertainty estimation. With neural networks often comprising millions or billions of parameters, the challenge lies in understanding effective optimization techniques for these networks. I am also exploring the phenomenon of why neural networks have such impressive generalization abilities. And finally, another key aspect of my research was investigating whether neural networks can possess awareness of their knowledge gaps.

How is your work with AI going to improve life for the average person?

My research on neural network training dynamics holds significant importance in the realm of large language models and AI research. These models used in programs such as ChatGPT, which have become ubiquitous in our daily lives, play a vital role in various applications. For example, they assist us with translation, enhance our essay writing and serve as virtual assistants to address our queries.

There’s controversy surrounding some of the ways AI is being used – why do you think people need to embrace its potential?

The public should embrace AI as a catalyst for positive change because it enhances efficiency and productivity across industries, automates mundane tasks allowing for more meaningful work and improves problem-solving and decision-making through data analysis. In addition, it augments human capabilities and drives innovation while also helping us to address societal challenges like inequality and sustainability.

Embracing AI responsibly ensures transparency, accountability and ethical considerations, unlocking AI's potential for positive impact in our society. I think the public should also be involved in regulating AI, as AI systems could be very powerful and misuse of them could lead to catastrophic consequences.

What career path will you pursue after graduation – and how will your U of T education help you excel?

I will work as an AI researcher in industry, focusing on large language models. My education at U of T has equipped me with extensive knowledge in deep learning and artificial intelligence. Under the guidance of my advisor, Associate Professor Roger Grosse, and collaboration with colleagues, I have gained valuable insights into neural network training dynamics and AI safety.

This expertise enables me to enhance the efficiency of large language model training while ensuring alignment with human values. My PhD work on understanding neural network training with a noisy quadratic model has already been used by many big industrial labs (including Anthropic, DeepMind, OpenAI) in training the latest models.

What advice do you have for people considering their PhD – in your field or beyond?

I recommend everyone maintain a curious mindset and pursue their passions. Curiosity is vital for scientific progress. It is also crucial to remain open-minded and committed to lifelong learning. Our field is rapidly evolving, rendering knowledge from just a few years ago potentially outdated, so continuous learning is essential.

Research shows decision-making AI could be made more accurate when judging humans

siddiq22 — Tue, 23 May 2023 22:08:43 +0000

Research shows decision-making AI could be made more accurate when judging humans

siddiq22 Tue, 05/23/2023 - 18:08

Cutline

(photo by wildpixel/iStock)

Jovana Jankovic

Alexander Bernier

Topic

Breaking Research

Schwartz Reisman Institute for Technology and Society

Artificial Intelligence

Equity

machine learning

Research & Innovation

Subheadline

New study by researchers from U of T and MIT suggests that clearly labelling data might help reduce bias

A new study from researchers at the ��Ƶ and the Massachusetts Institute of Technology (MIT) is challenging conventional wisdom on human-computer interaction and reducing bias in AI.

The paper, which was published this month in the journal Science Advances, demonstrates empirical evidence on the relationship between the methods used to label the data that trains machine learning (ML) models and the performance of those models when applying norms.

MIT PhD student Aparna Balagopalan, a graduate of U of T's masters program in applied computing, is lead author, with co-authors Gillian Hadfield, director of U of T’s Schwartz Reisman Institute for Technology (SRI), Schwartz Reisman Chair in Technology and Society, CIFAR AI Chair, and a professor of law and strategic management in the Faculty of Law; David Madras, a PhD student in the Machine Learning Group at the department of computer science in the Faculty of Arts & Science and the Vector Institute; research assistant David H. Yang, a graduate student in the applied computing program in the Faculty of Arts & Science; Marzyeh Ghassemi, a faculty affiliate at SRI and an assistant professor at MIT; and Dylan Hadfield-Menell, an assistant professor at MIT.

Much of the scholarship in this area presumes that calibrating AI behaviour to human conventions requires value-neutral, observational data from which AI can best reason toward sound normative conclusions. But the new research suggests that labels explicitly reflecting value judgments, rather than the facts used to reach those judgments, might yield ML models that assess rule adherence and rule violation in a manner that humans would deem acceptable.

To reach this conclusion, the authors conducted experiments to see how individuals behaved when asked to provide factual assessments as opposed to when asked to judge whether a rule had been followed.

From left to right: MIT PhD student Aparna Balagopalan, SRI Director Gillian Hadfield and SRI Faculty Affiliate Marzyeh Ghassemi (supplied photos)

For example, one group of participants was asked to label dogs that exhibited certain characteristics – namely, those that were large, not well groomed, or aggressive. Meanwhile, another group of participants was instead asked whether or not the dogs shown to them violate a building pet code predicated on the same characteristics, rather than assessing the presence or absence of specific features.

The first group was asked to make a factual assessment – and the second, a normative one.

Hadfield says the researchers were surprised by the findings.

“When you ask people a normative question, they answer it differently than when you ask them a factual question,” she says.

Human participants in the experiments were more likely to recognize (and label) a factual feature than the violation of an explicit rule predicated on the factual feature.

Current thinking on this topic presumes that calibrating AI behaviour to human conventions requires value-neutral, observational data from which AI can best reason toward sound normative conclusions.

But this new research suggests that labelling data with labels that explicitly reflect value judgments, rather than the facts used to reach those judgments, might yield ML models that assess rule-adherence and rule-violation in a manner that we humans would deem acceptable.

The results of these experiments showed that ML models trained on normative labels achieve higher accuracy in predicting human normative judgments. Essentially, they are better at predicting. Therefore, if we train automated judgment systems on factual labels – which is how several existing systems are being built – they are likely overpredicting rule violations.

The implications of the research are significant. Not only does it show that reasoning about norms is qualitatively different from reasoning about facts, but it also has important real-world ramifications.

“People could say, ‘I don’t want to be judged by a machine – I want to be judged by a human,’ given that we’ve got evidence to show that the machine will not judge them properly,” Hadfield says.

“Our research shows that this factor has a bigger effect [on an ML model’s performance] than things like model architecture, label noise and subsampling – factors that are often looked to for errors in prediction.”

Ensuring that the data used to train decision-making ML algorithms mirrors the results of human judgment – rather than simple factual observation – is no small feat. Proposed approaches include ensuring that the training data used to reproduce human assessments is collected in an appropriate context.

To this end, the authors of the paper recommend that the creators of trained models and of datasets supplement those products with clear descriptions of the approaches used to tag the data – taking special care to establish whether the tags relate to facts perceived or judgments applied.

“We need to train on and evaluate normative labels. We have to pay the money for normative labels, and probably for specific applications. We should be a lot better at documenting that labelling practice. Otherwise, it’s not a fair judgment system,” Hadfield says.

“There’s a ton more research we need to be doing on this.”

The study was funded by the Schwartz Reisman Institute for Technology and Society and the Vector Institute, among others.

Researchers develop interactive ‘Stargazer’ camera robot that can help film tutorial videos

siddiq22 — Fri, 19 May 2023 19:02:56 +0000

Researchers develop interactive ‘Stargazer’ camera robot that can help film tutorial videos

siddiq22 Fri, 05/19/2023 - 15:02

Cutline

Research led by U of T computer science PhD candidate Jiannan Li explores how an interactive camera robot can assist instructors and others in making how-to videos (photo by Matt Hintsa)

Topic

Our Community

Computer Science

Faculty of Arts & Science

Faculty of Information

machine learning

Research & Innovation

A group of computer scientists from the ��Ƶ wants to make it easier to film how-to videos.

The team of researchers have developed Stargazer, an interactive camera robot that helps university instructors and other content creators create engaging tutorial videos demonstrating physical skills.

For those without access to a cameraperson, Stargazer can capture dynamic instructional videos and address the constraints of working with static cameras.

“The robot is there to help humans, but not to replace humans,” explains lead researcher Jiannan Li, a PhD candidate in U of T's department of computer science in the Faculty of Arts & Science.

“The instructors are here to teach. The robot’s role is to help with filming – the heavy-lifting work.”

The Stargazer work is outlined in a published paper presented this year at the Association for Computing Machinery Conference on Human Factors in Computing Systems, a leading international conference in human-computer interaction.

Li’s co-authors include fellow members of U of T's Dynamic Graphics Project (dgp) lab: postdoctoral researcher Mauricio Sousa, PhD students Karthik Mahadevan and Bryan Wang, Professor Ravin Balakrishnan and Associate Professor Tovi Grossman; as well as Associate Professor Anthony Tang (cross-appointed with the Faculty of Information); recent U of T Faculty of Information graduates Paula Akemi Aoyaui and Nicole Yu; and third-year computer engineering student Angela Yang.

A study participant uses the interactive camera robot Stargazer to record a how-to video on skateboard maintenance (supplied photo)

Stargazer uses a single camera on a robot arm, with seven independent motors that can move along with the video subject by autonomously tracking regions of interest. The system’s camera behaviours can be adjusted based on subtle cues from instructors, such as body movements, gestures and speech that are detected by the prototype’s sensors.

The instructor’s voice is recorded with a wireless microphone and sent to Microsoft Azure Speech-to-Text, a speech-recognition software. The transcribed text, along with a custom prompt, is then sent to the GPT-3 program, a large language model which labels the instructor’s intention for the camera – such as a standard versus high angle and normal versus tighter framing.

These camera control commands are cues naturally used by instructors to guide the attention of their audience and are not disruptive to instruction delivery, the researchers say.

For example, the instructor can have Stargazer adjust its view to look at each of the tools they will be using during a tutorial by pointing to each one, prompting the camera to pan around. The instructor can also say to viewers, "If you look at how I put ‘A’ into ‘B’ from the top,” Stargazer will respond by framing the action with a high angle to give the audience a better view.

In designing the interaction vocabulary, the team wanted to identify signals that are subtle and avoid the need for the instructor to communicate separately to the robot while speaking to their students or audience.

“The goal is to have the robot understand in real time what kind of shot the instructor wants," Li says. "The important part of this goal is that we want these vocabularies to be non-disruptive. It should feel like they fit into the tutorial."

Stargazer’s abilities were put to the test in a study involving six instructors, each teaching a distinct skill to create dynamic tutorial videos.

Using the robot, they were able to produce videos demonstrating physical tasks on a diverse range of subjects, from skateboard maintenance to interactive sculpture-making and setting up virtual-reality headsets, while relying on the robot for subject tracking, camera framing and camera angle combinations.

The participants were each given a practice session and completed their tutorials within two takes. The researchers reported all of the participants were able to create videos without needing any additional controls than what was provided by the robotic camera and were satisfied with the quality of the videos produced.

While Stargazer’s range of camera positions is sufficient for tabletop activities, the team is interested in exploring the potential of camera drones and robots on wheels to help with filming tasks in larger environments from a wider variety of angles.

They also found some study participants attempted to trigger object shots by giving or showing objects to the camera, which were not among the cues that Stargazer currently recognizes. Future research could investigate methods to detect diverse and subtle intents by combining simultaneous signals from an instructor’s gaze, posture and speech, which Li says is a long-term goal the team is making progress on.

While the team presents Stargazer as an option for those who do not have access to professional film crews, the researchers admit the robotic camera prototype relies on an expensive robot arm and a suite of external sensors. Li notes, however, that the Stargazer concept is not necessarily limited by costly technology.

“I think there’s a real market for robotic filming equipment, even at the consumer level. Stargazer is expanding that realm, but looking farther ahead with a bit more autonomy and a little bit more interaction. So realistically, it could be available to consumers,” he says.

Li says the team is excited by the possibilities Stargazer presents for greater human-robot collaboration.

“For robots to work together with humans, the key is for robots to understand humans better. Here, we are looking at these vocabularies, these typically human communication behaviours,” he explains.

“We hope to inspire others to look at understanding how humans communicate ... and how robots can pick that up and have the proper reaction, like assistive behaviours."

Add new author/reporter

Krystle Hewitt

Researcher uses ‘fuzzy’ AI algorithms to aid people with memory loss

Christopher.Sorensen — Thu, 14 Jul 2022 14:06:32 +0000

Researcher uses ‘fuzzy’ AI algorithms to aid people with memory loss

Christopher.Sorensen Thu, 07/14/2022 - 10:06

Cutline

(Image by Eugene Mymrin via Getty Images)

Matthew Tierney

Topic

Our Community

Artificial Intelligence

Brain

Electrical & Computer Engineering

Faculty of Applied Science & Engineering

machine learning

Memory

A new computer algorithm developed by the ��Ƶ’s Parham Aarabi can store and recall information strategically – just like our brains.

The associate professor in the Edward S. Rogers Sr. department of electrical and computer engineering, in the Faculty of Applied Science & Engineering, has also created an experimental tool that leverages the new algorithm to help people with memory loss.

“Most people think of AI as more robot than human,” says Aarabi, whose framework is explored in a paper being presented this week at the IEEE Engineering in Medicine and Biology Society Conference in Glasgow. “I think that needs to change.”

Parham Aarabi

In the past, computers have relied on their users to tell them exactly what information to store. But with the rise of artificial intelligence (AI) techniques such as deep learning and neural nets, there has been a move toward “fuzzier” approaches.

“Ten years ago, computing was all about absolutes,” says Aarabi. “CPUs processed and stored memory data in an exact way to make binary decisions. There was no ambiguity.

“Now we want our computers to make approximate conclusions and guess percentages. We want an image processor to tell us, for example, that there’s a 10 per cent chance a picture contains a car and a 40 per cent chance that it contains a pedestrian.”

Aarabi has extended this same fuzzy approach to storing and retrieving information by copying several properties that help humans determine what to remember — and, just as critically, what to forget.

Studies have shown that we tend to prioritize more recent events over less recent ones. We also emphasize memories that are more important to us and we compress long narratives to their essentials.

“For example, today I remember that I saw my daughter off to school, I made a promise that I’d pay someone back and I promised that I’d read a research paper,” says Aarabi. “But I don’t remember every single second of what I experienced.”

The capacity to overlook certain information could supercharge existing models of machine learning.

Today, machine learning algorithms trawl through millions of database entries, looking for patterns that will help them correctly associate a given input with a given output. Only after countless iterations does the algorithm eventually become accurate enough to deal with new problems that it hasn’t already seen.

If bio-inspired artificial memory enables these algorithms to give prominence to the most relevant data, they could potentially arrive at meaningful results much more quickly.

The approach could also support tools that process natural language to help people with memory loss keep track of key information.

Aarabi and his team have set up such a tool using a simple email-based interface. It reminds participants of important information based on algorithmic priority and a relevant index of keywords.

“Ultimately, it’s geared to people with memory loss,” Aarabi says. “It helps them remember things in a way that’s very human, very soft, without overwhelming them. Most task management aids are too complicated and not useful in these circumstances.”

The demo is free and available for anyone to play with; simply send an email to mem@roya.vc for instructions.

“I’ve been using it myself,” says Aarabi. “The goal is to put the demo in people’s hands – whether they’re dealing with significant memory degradation or just everyday pressures – and see what feedback we get. The next steps would be to build partnerships in health care to test in a more comprehensive way.”

“These days, AI applications are increasingly found in many human-centred fields,” says Professor Deepa Kundur, chair of the department of electrical and computer engineering. “Professor Aarabi, by researching ways to better integrate AI with these ‘softer’ areas, is looking to ensure that the potential of AI is fully realized in our society.”

Aarabi says that this algorithm is just the beginning.

“Biologically inspired memory may very well take AI a step closer to human-level capabilities.”

machine learning

Space doctor: U of T researcher co-chairs International Space University program in Brazil

Read more about Farhar Asrar’s research at U of T News

Researchers accelerate development of cellular therapies for damaged tissues

New autonomous lab at ������Ƶ to improve drug formulation

Researchers explore shifting AI landscape at Absolutely Interdisciplinary conference

Cognition, storytelling and the future of intelligence

AI, learning and the role of education

AI’s capabilities, behaviours and harms

New languages, new frameworks for understanding

Read the full story at the Schwartz Reisman Institute

How AI and neuromodulation could help with sleep disorders

From machine learning to mentorship, graduate Irene Fang showed leadership during her time at U of T

New PhD graduate Guodong Zhang hopes to lead the next generation of AI experts

Research shows decision-making AI could be made more accurate when judging humans

Researchers develop interactive ‘Stargazer’ camera robot that can help film tutorial videos

Researcher uses ‘fuzzy’ AI algorithms to aid people with memory loss

New autonomous lab at ��Ƶ to improve drug formulation