 It's my great pleasure to introduce our associate professor, Joaquin Gony Cortez, who joined Purdue in 2015 as a faculty member in industrial engineering, biomedical engineering, and Purdue Systems Collaboratory. So you can already tell that Dr. Gony is a highly interdisciplinary in his research work. So he received the PhD in physics and applied the math in the University of Nevada, Spain. And then he did his postdoc actually at the applied medical research lab at the same university. And then also he had a postdoc and research opportunities at IU, Indiana University in multiple positions in psychological and brain sciences and network science institute. So you will hear more about his research, but he does truly interesting outstanding work, try to understand brain structure and functions to see how they can affect our disease and behavior and cognition. Okay, so he has a journal publication in this area, 69 in top journals, and his citations are about 5,000, which is very strong in his career right now. So he received the multiple awards Purdue Discovery Park Data Science Award, and he received the undergraduate educator award Pritchaker, which is a very competitive in school of IE. And he received the two times, including last year. And then he received the college of the teaching award 2021. Among his many education contributions, one of the notable one is called the May master. Okay, May master is a Spain, a broader program, and students go there for about a month. And they actually take courses in Spanish and system dynamics from Dr. Goni, and they eat together, they visit, you know, the industry, and then, you know, have some fun together. And I had a chance to talk with the, I see Rachel Snow there, who is our first, who received the first place of the undergraduate poster presentation last week. And when I talk with Rachel, she was very excited about this upcoming May master. And I see also Morgan, our undergraduate advisor here, who will also join the May master group in May. Okay, so let's welcome Dr. Goni Cortez to the podium. Good afternoon, everyone. Is this working? Okay, it's a pleasure for me to be here. I'm going to go through what I titled as a few highlights on my academic journey. I have to say it was really difficult to pick what to put and what not. I added the slides, I removed them, I went back and forth. So I'm going to do my best to stick to time. A little bit about me. I'm going to take Dr. Son, beautiful introduction. Thank you so much for it, to skip a little bit on my academic background. I had a postdoctoral experience in Spain right after my PhD, a second postdoctoral experience in Indiana University, research scientist position, and then finally I joined Purdue in August 2015. A little bit more about my geographical historical origins. I am from Pamplona. Pamplona is the capital of Navarra, which is an autonomous community in Spain. A couple of links that may trigger your interest. Yes, Pamplona is where it happens, the running of the bulls. And no, I have never done it. And you also might know there is a strong connection between Pamplona and the U.S. thanks to the famous writer Ernest Hemingway, who actually wrote a book about the festivals in Pamplona and will honor us with his presence for literally decades as the festivals were happening every July. I want to start by my lab. After a few months, I had joined Purdue. I started the complexity lab, it's not a typo. It's a fusion of connectomics and complexity with an M. I was remembering when I was doing this slide that at the beginning you would type in Google complexity and no hits were found. And then as we started publishing papers and getting to be known, now if you type complexity, it's not trying to correct the word to complexity, but actually you should find our research lab website. Our current members are Mintao Liu, Daniel Guerrero, and Maddie Mohadam, who are here with us, undergraduate student Rachel Snow. And I don't want to forget our fantastic alumni, including Dr. Enrico Amico, who is now in EPFL in Lausanne as a principal investigator, Minusri Rajapandyan, who is now used to be in gecko robotics, but I know that she has moved to another company, but she keeps being extremely successful in her career. We also have Michael Wang, who is in consulting, Uttara Thymnis, who is in the Livermore National Livermore Labs. That was a great experience through a collaboration that was only possible because BME, our wilderness school of biomedical engineering, took as an initiative to Livermore Labs to try to foster collaborations. Not only we did a collaboration, but actually Uttara ended up after earning her PhD working there. We have Kosar Abbas, who is now in industry. He is a senior researcher at Intel, and we have Dewey Duontran, who just started six months ago, his position as an assistant professor at the Naval Academy Department of Mathematics. I'm going to go very little, and this was a strong struggle what to choose and what not. Through the research I have done in my career, I want to highlight a little bit on the concept of network morpho spaces where maybe some of our work was relatively pioneering the field. So mapping a network morpho space reveals the extent to which the space is filled by existing networks. That's allowing a distinction between actual and impossible designs and highlighting the generative potential of rules and constraints that pervade the evolution of complex systems. For this topic of network morpho spaces, what it's trying to do is to build a really low dimensional space to three coordinates maximum, in which we can understand any complex system that can be represented as a network or graph, which is nothing but a set of nodes and a set of edges or links connecting those nodes. In this paper, we tried to tackle the concept of hierarchy. How can we determine or quantify hierarchy in complex networks with the additional challenge of dealing with directed graphs, which could be a C-click where you have no cycles A to B, B to C, C to A, for instance, but also that they could have any possible size, degree, just a universal set of mathematically principled measurements in which every network, any directed graph will have a set of three coordinates which we named treeness, fear forwardness, excuse me, fear forwardness, and orderability. Then we went through the conceptual aspect of what are possible coordinates in this morpho space and what are not. Can different graphs of any size or degree or other characteristics filled entire three dimensional morpho space? The answer is no. There are impossible regions that cannot satisfy certain level of treeness, orderability, and fear forwardness. Within the possible, what's possible from a graph theoretical perspective, where are the real systems living in this space? Do they occupy the entire space that is possible or actually systems that come from data from real systems occupy specific regions? The specific regions of the morpho space that are occupied, are they theme related? Are language networks occupying certain type of the space and energy communication networks occupying another space and food webs occupying a different portion of the space? The answer to all those is yes. There is a specificity in the location. I'm going to skip the details and I'm going to the next concept. We also designed a two dimensional communication morpho space in which the x-axis is how efficient it is to explore a network when you are not aware of the structure or topology of the network. So you don't know, you are in a node. You want to reach another node, but you don't know what the shortest path. On the y-axis, on the contrary, is for the same network as a second coordinate, which is when I am in a complex network and I am in a node and I want to reach another node, how fast is to travel to the other node given that I have full topological information of the system. Think about we're typing the GPS coordinates of a destination that we want, that will be the parallelism to a shortest path. And then we again played with the idea of the possible and the impossible. Where are any possible graphs, in this case of 50 nodes and average degree of 4, which means that every node is connected on average to 4 over neighbors. What is possible and what is not in this morpho space? And then we tried to see, okay, now that we find, let me call it the perimeter on this morpho space of what is possible and what not, let's look how those networks look like. And it was really fun and it was triggering me a lot of memories doing these slides. The worst possible network in terms of communication and division, where you are not aware of the structure or in routing when you are actually aware of the structure. This was obtained computationally, but this is brilliant. This is a terrible network to navigate through. This is a highly connected cluster with a very long corridor to another highly connected cluster. Any communication between cluster A and cluster B, even if you know the shortest path, is just horrible. It's really bad. And we discovered all these topologies that go all the way to the one that is maximally efficient for both in a computational Pareto front multi-objective simulation. Then as part of my career, I was like, okay, I love complex systems. I love theoretical and simulation aspects of complex systems. Where could I apply this to learn more and to see what kind of domains could benefit from that? And then I enter into this fascinating world of brain connectomics. In brain connectomics, long story short, you have different brain regions in the cortex from which you produce, sorry, you don't produce. You estimate, you measure time series which represent neural activity with respect to time. And then when you do the per-wise correlations between all those time series, you end up with an object that from a mathematical standpoint is nothing but a correlation matrix. From a brain connectomics standpoint, it's called functional connectome. And it tells me for any pair of two brain regions to what extent they are functionally coupled. If you want for that subject doing whichever task the subject is doing, to what extent there is communication between those two brain regions. Now, automatically, you should be getting excited and thinking, well, this functional connectome is an object that is alive. My functional connectome right now, when I was having breakfast this morning or when I was sleeping, not that much last night, is different. And then also, when subjects are doing the same task, it could be that different subjects have different characteristics in the functional capital of the brain regions. So even when different subjects do the same task, there might be some differences. And this is how the concept of fingerprinting emerged of to what extent we can differentiate subjects, for instance, just to make it simple when they are all doing the same task. So, intellectually, it was incredibly challenging to me. And the game that is called identification rate is the percentage of times, so the value is 0 to 1, in which I give you a functional connectome of a hidden participant X. And you need to figure out this functional connectome belongs to who in a set of functional connectomes in which I know who the person is. The rational is you compare the functional connectome to each and all the other functional connectomes, and either through a concept of distance or similarity, whoever scored the highest, excuse me, the highest, is your guess or your prediction of who is the subject. The percentage of times that you are successful in this process is called identification rate or also simply accuracy. We, in collaboration with the group of Luis Pessoa, developed a framework in which we try to be aware on the data characteristics in our way to measure similarity or distance in the functional connectomes. If you take two correlation matrices and remember functional connectomes are nothing but two correlation matrices, there are dozens of ways in which you can easily think you could measure either some similarity score or some distance score. But I'm going to use the brain actually as an example of things that are more principled than others. If I want to know on the perimeter, on the surface of the cortex, what is the distance between this point and this point, I could cheat and measure Euclidean distance. Except when you're measuring Euclidean distance, you're not really preserving or being aware of the geometry of your object. So yes, you're going to get a value, but that value is not truly representative of the distance between those two points. This is what we brought together with the group of Luis Pessoa as measuring geodesic distance, which is geometrically aware of the n-dimensional space where all the correlation matrices of that size occupy. So when doing that, Pessoa first and colleagues showed that accuracy identification rate percentage of times that actually reconcile functional connectome of a subject that you don't know who it is to the correct subject systematically is higher when you actually use geodesic distance. Geodesic distance has a tricky challenge sometimes because geodesic distance needs the inverse of our correlation matrices. I'll make it very short and not painful, but for a functional connectome represented by a correlation matrix to be invertible, I need more time points in my time series measurements than brain regions I'm using in my parcelation of the cortex. That means that if my task is relatively short, I cannot measure geodesic distance unless you do a concept that is called regularization. They use a fixed regularization on one, and we extended collaborating with them. It was a really fruitful collaboration that the amount of regularization that you do in your matrices matter as of what is the consequences in terms of the fingerprint, and we saw that actually optimal regularization to uncover fingerprints depends on the scanning length, on the brain parcelation, and on the task that you're looking at. We kept forward and I got absolutely obsessed with Riemannian geometry in which now you can project your correlation matrices to something that is called tangent space in which finally the Euclidean distance in that space is equivalent to the underlying geodesic distance in the manifold where all the correlation matrices leave. If you examine the metaphor, it's actually fairly similar or at least it has some strong similarities with the concept of building a two-dimensional Alice from in this case planet Earth. You need to choose a reference point and then you do a projection where you try to the best that you can. There are distortion issues that the Euclidean distance between any two points represent, truly represent the distance that was happening in the original object. So by doing that, and I'm going to skip some technical details, we actually saw that you can reach unprecedented levels of fingerprint on the same data set that were tested before. The only thing we're doing is a tangent space projection and we are seeing that for any task, any scanning length. Not only test, retest, but now I take one monocyclic twin and I try to find who is the other monocyclic twin or I take one di-cyclic twin and I try to figure out who is the other di-cyclic twin. We reach unprecedented levels of fingerprinting with accepting the challenge that in di-cyclic twins they serve as much genetic material as simply siblings. We are very far from getting 100%. We saw this for any fMRI task, for any brain parcelation, for any scanning length and for the entire fingerprint gradient, not only test, retest. As Son was mentioning, I'm very excited to share that the Pamplona study abroad in Pamplona that I led in 2017 and 2018. Thanks to the really valuable support of Dr. Son, we are back to it in 2023. And I'm taking this slide to start saying my thank you slides, where I really want to thank Alfonso Larioth, who is the director of the Pamplona Learning Institute. Miquel Galvaris, who is one of the teachers that will teach our undergrad Spanish. Elisabeth Pearson, who was my partner in 2017 and 2018 on this trip. And Morgan Curilla, who is going to be my partner in a month from now. So thank you all for making this possible and thank you Dr. Son for supporting the program so much. I needed this statement. There are many people to thank you for my academic career that I will not be giving this presentation without their support, trust, empathy, and work. And I'm sure I'm forgetting a few. But I want to highlight Professor Olaf Espons, who brought me to the U.S. and trusted me when I was just an interesting, passionate scientist that didn't know much about anything. And he trusted me and I was three years post-doc with him. Professor Ricard Soleil, one of the greatest minds in complex systems. And during my PhD, whenever I had a chance, I would escape to Barcelona for a week to work with him and his group, including Bernat Corominas-Murtra, Carlos Rodríguez Caso, and Sergi Valverde. My probably best collaborators from IU School of Medicine, Professor David Carrigan, Mario Chemichic, and also from Epidemiology and Biostatistics, Jaroslav Hareslak. This is a great team in which I felt so supported ever since I joined Purdue. And we have great collaborations going on. My colleagues in the department, Heswaldo Scutari, URNG, Mario Ventresca, they were great colleagues, mentors, and whenever I needed something, they were always there for me. And it's something that I will do my best to propagate to our new incoming assistant professors. And in terms of leadership, Steve Landry and Abby Desmuth, and our not new but relatively recent son, who have been so supportive to my career and my research. And of course, Dr. Patrick Brunez, who has been a pleasure to work with in all kinds of things, from supporting the main master to helping me coordinate all the courses that we teach. Big thank you to my team. There is nothing I could do without my team, and I'm really grateful. So thank you very much. And there is nothing I could do without my family. My family is distributed into three countries. I have my wife, Diana, our son, Santiago, seven years old, six and a half, and he wants to be a scientist apparently. And our family is distributed in Spain, where is my mother, Teresa, my sister, Maria, in law, Miguel, and nephew and niece, Miguel and Teresa. We also have a branch in Colombia through my wives, where we have a lot of cousins and uncles and aunts. And Diana's parents are also here in Indiana, Clara and Fernando, and they are all very supportive for me and my career. And I don't want to forget two dogs that represent a lot in my life, Yalita, who left us last year, and Mima, who is still enjoying the life with us. And thank you so much. Great presentation. So I forgot to mention that... Hello. So Rachel, I mentioned Rachel and she was on the photo and she was advised by Joaquin. That's why another reason that I mentioned. And also main master program, it looked, I mean you may have noticed, the program looked so interesting. I actually almost wanted to participate, but since I'm moving, I'm moving from actually, my family is moving, and you know, I cannot join it, but again, so thanks to all your contribution. So any question for Dr. Goni Cortez? I told you that he does very, very exciting the research work, right? So, okay, well let me start with... Yeah, thank you for sharing your fantastic work. That was very insightful. And I guess, you know, there is a piece of for looking forward, right? So what is your research vision in the next five or ten years, and how College of Engineering can continue to support the law of the researcher already doing so fantastic? So the next steps in the lab, I want to put more emphasis, more effort into longer projects where, for instance, we look at longitudinal data, how a functional connectome of a subject changes across time, less thing, for instance, simply aging, but also because of a disease progression, but also because of a therapy, or also because of a drug that is a treatment for certain disease. This needs to collect data for years, and it's that we make small steps in our computational models as the project acquires more and more data. And now that I'm an associate professor, I think I have the calm, if that makes sense, that we can pursue these projects that could be four, seven, eight years of continuous work until we truly get the value of how the functional connectomes evolve in the long term. I also have, in a shorter term strategy, a lot of interest in quick functional reconfigurations. What happens when a subject is finishing a task? Let's say playing chess. And now I tell you, okay, we have finished the game. Now we go to rest. Obviously, we don't go to rest. We're still thinking about that move or why I didn't see that queen there, or that king. So the way we go back to rest after a highly engaging task, that was my metaphor for playing chess, is different across subjects, and it's something that I'm extremely interested to keep digging into. Any other questions? Hi, Joaquin. He's my husband. This is not planned. I'm actually shaking right now. Oh, something that I have always found very unique and interesting about your career is that you have worked in an extremely interdisciplinary fashion. So you have your undergrad in, I think, informatics or computer engineering, phd in mathematics, then you worked in psychology and network science, and now you're a professor in industrial engineering. And so I was just wondering what, and I think that that's pretty rare, but hopefully will become more common in the space of engineering and research and science. What characteristics do you think about you have kind of allowed you to navigate that so well, and kind of what vision do you have for the world to become more collaborative? So as you probably noticed, I am highly collaborative in interdisciplinary environments. I think it's part of my personality trait and my passion for science. I don't so much put labels, oh, this part of the world belongs to this category or this type of engineering or a psychologist or a neurologist. Obviously, there are some parts that are. But what I truly like on interdisciplinary projects is when we really work together and we set the questions that we want to understand and we set the modeling or the mathematical approaches that we're going to use to answer those questions, and then we just go and proceed. And it's something that I give a lot of importance to in a principled manner. I don't think this is good advice. I didn't have a clear plan when I finished my undergrad studies or even when I pursued my PhD. It was purely coming from my passion for science and every step took me to the other. And I think I probably overcame my lack of planning through my passion for science. Any other questions for Joaquin? Okay, Joaquin, great presentation and look forward to continuing to see your continued success next few years. Okay, let's give him another round of applause.