Enhancing Human-Machine Interaction!
The field of speech quality measurement has undergone remarkable advancements, driven by the continuous evolution of telecommunications and neuro-engineering. The industry has broadened its scope to encompass various human-machine interactions, paving the way for innovative technologies that enhance communication and user experiences. As research pushes the boundaries of artificial intelligence (AI), machine learning, and signal processing, the applications within this field continue to grow, with implications ranging from healthcare to cybersecurity. Researchers are now exploring how these advancements can impact multisensory experiences, bridging the gap between virtual and physical realms, which holds great potential for the future of immersive technologies.
Tiago H. Falk, a prominent figure in this field, has demonstrated exceptional leadership in shaping the direction of speech quality measurement and neuro-engineering. As a Professor and Director, his innovative approach has helped redefine how machines interact with human senses. Tiago’s vision and strategic thinking have influenced his team and the wider research community. His ability to bring diverse perspectives together cultivates advanced research, driving progress in applied AI, signal processing, and multisensory applications.
At the Institut national de la recherche scientifique (INRS), part of the University of Québec, Tiago leads the Multisensory Signal Analysis and Enhancement Lab. This lab focuses on developing modern technologies that integrate multiple senses to improve virtual reality, healthcare, and AI applications. The institute has built a strong reputation for producing world-class research in neuroengineering and telecommunications, collaborating with multinational companies to address societal challenges. Through its innovative work, INRS is at the head of innovation, contributing to significant advancements in academic research and practical applications in the tech industry.
Let’s explore Tiago’s inventive approach in neuroengineering research:
Advancements in Speech Quality Measurement
Tiago obtained his BSc degree in Electrical and Computer Engineering from the Federal University of Pernambuco, Brazil, in 2002. Soon after graduation, he moved to Canada for graduate studies at Queen’s University in Kingston, Ontario. By the end of 2008, he had obtained his MSc and PhD in the topic of speech quality measurement, where a model of how humans perceive the quality of a phone call was developed.
Upon graduating, he started a postdoctoral fellowship at the University of Toronto’s Holland Bloorview Children’s Rehabilitation Hospital, developing assistive technologies for non-verbal children with multiple severe disabilities. This is when he ventured into neurotechnologies and neuroengineering to develop brain-machine interfaces.
In 2010, he was hired at the Institut national de la recherche scientifique (INRS), a graduate-level university, part of the University of Quebec network, as an Assistant Professor. His goal was to combine neuroengineering with telecommunications to develop next-generation human-machine interface tools.
He founded the Multimodal/Multimedia Signal Analysis and Enhancement (MuSAE) Lab to achieve this goal. Fourteen years later, he is now a Full Professor and co-director of the INRS-UQO Joint Research Unit on Cybersecurity. The MuSAE Lab was rebranded as the Multisensory Signal Analysis and Enhancement Lab in 2018 to align with emerging research interests in virtual reality, metaverse, and, more recently, the Internet of Senses.
Leading a Diverse Research Team
He currently leads a team of two postdoctoral fellows, eight PhD students, one MSc student, and an undergraduate intern. Each member is working on a separate project, usually in partnership with Canadian and/or multinational companies, to address issues of relevance to society.
These projects range from using edge computing, machine learning, responsible AI, and signal processing for different applications, such as automated beehive monitoring, deepfake detection, multisensory experiences for next-generation gaming and healthcare, authentication and access control methods for the metaverse, and secure and robust speech applications, among others.
As a Professor, Tiago is also involved in teaching various graduate courses focusing on vulnerabilities of AI systems, multimodal human-machine interfaces, and signal processing for biomedical applications. Additionally, his role involves delivering services to the university and the research community in general. These services include sitting on committees, organizing conferences, serving on editorial boards, and reviewing papers.
Effective Time Management Strategies
Tiago believes balancing dual roles as a Professor and Co-Director effectively requires discipline. While many roles as co-directors are aligned with the roles of a professor (such as searching for partnerships with companies and international universities, writing grant proposals, publishing, and showcasing the work of students), administrative tasks can take some time. For this, he allocates several blocks in his calendar during the week to focus 100% on these tasks. Sticking to this routine has been essential in balancing both roles.
Encouraging Diverse Perspectives in Research
His team is very interdisciplinary, comprising individuals with backgrounds in engineering, computer science, psychology, and physics. These diverse experiences allow for different views and ideas to be brought to the table. When leading various projects, he ensures everyone has the time and space to share their ideas and discuss their potential usefulness for the project.
For example, it is customary to have weekly, biweekly, or monthly meetings where everyone discusses their progress and challenges and shares ideas on how to overcome them. These interactions teach students how to make decisions independently, be open to different ideas, and integrate concepts from other fields into their own work. These are essential elements in becoming independent researchers and scientists, benefiting everyone involved.
In the lab, they have worked with the same partners for many years. In one case, they have been collaborating with the same partner for over a decade and have been awarded numerous prizes together.
Funding and Partnership Strategies
Tiago finds that challenges in research evolve based on the academic’s career level. As a younger professor, his main challenge was attracting funding and partnerships to carry out the research. Today, the greatest challenge is student recruitment to carry out the funded research.
Working in areas such as applied AI, metaverse, and cybersecurity means competing with the giants of the world—Google, Meta, IBM, and Microsoft—with much deeper pockets than academia. Recruiting local students has become extremely challenging. In the past, international students were attracted, but with remote work enabled by the pandemic, many have also been drawn by the attractive paychecks from industry.
Overcoming this challenge is still a work in progress, and he has tried several strategies, from raising student scholarship amounts to giving bonuses based on productivity to helping students obtain summer internships at these corporations to going on “missions” to different countries and universities to set up partnerships. All these efforts have helped somewhat, but the problem is generalized and global.
Effective Work-Life Balance Strategies
He manages work-life balance with the same discipline used to manage multiple roles. Tiago relies heavily on his calendar, allocating dedicated slots for weekly meetings with students, administrative tasks, meetings with partners, service to the community and the university, and family activities. Unless there is an important imminent deadline, his weekends are dedicated to family, as are his evenings, at least until the kids’ bedtimes.
Staying Updated with Applied AI and Signal Processing
Tiago employs several strategies to stay updated with the latest developments in applied AI and signal processing, including:
- He attends several key conferences throughout the year to learn first-hand from the developers about these developments.
- Through his classes and weekly meetings with his students, he receives firsthand feedback from them on the latest innovations they have encountered.
- He actively uses LinkedIn and X to follow key individuals in his fields of interest to stay informed about their activities.
Essential Advice for Aspiring AI Researchers
He advises aspiring researchers and professionals in the field of applied AI to avoid merely applying others’ codes found online. He observes that many students and researchers fall into this practice. While open codes, datasets, and tools like LLMs and ChatGPT have democratized research, they have also led to a focus on marginal performance improvements without understanding the underlying principles.
To succeed in applied AI, one must possess domain knowledge and use it to enhance available open-source codes. This requires a solid understanding of the field, machine learning principles, signal processing, and statistics. Applied AI is inherently challenging, with noisy data and labels. It is crucial to learn how to examine, understand, and clean data before analysis.
Falk shares an example from his recent research, where data was collected over several weeks from hundreds of nurses using wearable devices to predict nurse burnout. Initial attempts to use a large neural network on this “big data” yielded results no better than random chance.
However, by improving data quality, extracting context with domain knowledge, and using relevant parameters in a classical machine learning algorithm, the performance improved markedly. Mastering these skills will help individuals stand out in the field of applied AI.
Expanding the Metaverse Beyond Audio and Visual
Tiago states that the metaverse today is mostly limited to audio and visual experiences via a head-mounted display and headphones. To create experiences that seamlessly blend real and virtual elements, it is necessary to stimulate other senses, including olfactory, haptic, and taste. The Internet of Senses (IoS) aims to develop the technologies needed to create, transmit, and render multiple senses in real-time.
In the MuSAE Lab, efforts are underway to develop IoS applications for the betterment of humankind. For example, IoS-enabled nature walks have been shown to alleviate stress in nurses, reduce symptoms of post-traumatic stress disorder patients, and aid in the neurorehabilitation of post-stroke patients. In gaming, IoS is being explored as a tool to reduce cybersickness and enhance the gaming experience.
For training, incorporating smells and heat may improve the training of first responders, such as police officers and firefighters. In urban planning, IoS could help build more empathic neighborhoods. Additionally, IoS is being explored to assist with the socialization of seniors in nursing homes, enabling them to be “present” at events like Thanksgiving dinners with their loved ones. There is much work to be done for IoS to ensure it will remain a focus for many years to come.
Future Research Goals for the Internet of Senses
Tiago outlines his research goals for the next 10 years. He aims to develop the necessary signal processing, machine learning, and cybersecurity tools to enable the widespread deployment of the Internet of Senses (IoS).
This will also require innovations in AI model compression, edge computing, and green and sustainable AI to ensure equitable access for everyone while protecting the planet.
Collaborations with colleagues in photonics have already begun, resulting in the development of an all-optical neural network capable of achieving promising results with just a fraction of the energy needed compared to its GPU counterpart and with additional cybersecurity benefits, such as robustness to adversarial attacks. This work was recognized as a 2023 Top 10 Discovery by the Québec Science Magazine.
