Double the Happiness: Two PhD students nab coveted Facebook Fellowships
Tue 04.28.20 / Miranda Adkins
Double the Happiness: Two PhD students nab coveted Facebook Fellowships
Tue 04.28.20 / Miranda Adkins
Tue 04.28.20 / Miranda Adkins
Tue 04.28.20 / Miranda Adkins
Double the Happiness: Two PhD students nab coveted Facebook Fellowships
Tue 04.28.20 / Miranda Adkins
Double the Happiness: Two PhD students nab coveted Facebook Fellowships
Tue 04.28.20 / Miranda Adkins
Tue 04.28.20 / Miranda Adkins
Tue 04.28.20 / Miranda Adkins
Recently, when the 36 winners of the 2020 Facebook Fellows were announced, and Lucianna “Lulu” Kiffer and Lydia Zakynthinou learned that they each had been awarded one of the prestigious fellowship, they were thrilled. Says Zakynthinou, “It was more than double the happiness.” Both are Khoury College PhD students and colleagues.
While they work in the same lab space on ISEC’s sixth floor, Kiffer and Zakynthinou’s research interests differ dramatically: Kiffer’s work focuses on blockchain protocols, and Zakynthinou’s research, more theory-based, focuses on machine learning and data privacy.
Both students learned of the Facebook Fellowship program, which makes awards to domestic and international students, through an email to Ph.D. students sent by Associate Dean and Professor Frank Tip listing upcoming fellowship and grant opportunities.
The fellowship award includes two years of tuition and fee coverage, an annual stipend, reimbursement for travel expenses, and funding to attend a summit for all 36 members of the fellowship cohort. “Facebook is very into the social part of the fellowship as well as the academic side,” Zakynthinou explains. “They encourage the fellows to reach out to each other and get to know each other.” She has already heard from a handful of other 2020 Facebook Fellows from different universities and research fields across the U.S. and abroad.
The purpose of the fellowship, according to the Facebook Research website, is to “encourage and support promising doctoral students who are engaged in innovative and relevant research.” Fellows have wide-ranging research interests that they are free to continue to pursue for the duration of the fellowship.
Kiffer’s current research on blockchain and cryptocurrency began during early months of her Ph.D. “One of my advisors, Alan Mislove, asked me to read up on this thing, Ethereum, and tell him how it works,” she recalls. “Ethereum is a cryptocurrency. I’d heard about Bitcoin before but never really read much about cryptocurrencies, but I just never stopped. It became my whole Ph.D.”
She believes that the potential of blockchain technology is far-reaching.
“Blockchain is still in its infancy, and there’s a lot to be improved,” she said. “ Even as a currency, Bitcoin has gone where nobody thought it was going to go. There are researchers looking at how to use Bitcoin or similar networks to do more than just currency — things like streaming services with micropayments on peer-to-peer networks, or distributed data storage on personal/work computers. The ability to have a system like this that anybody in the world can opt into is really powerful.”
Zakynthinou, on the other hand, is taking a more theory-based approach to her research, focusing mainly on Differential Privacy and Generalization. Both, she explains, are key aspects of machine learning theory.
“Differential Privacy is a mathematically rigorous technique that has become the standard for ensuring privacy in machine learning models,” Zakynthinou says. “It is based on the idea that if you change one data point in your sample set, then the distribution of the output of the algorithm should not change too much.” She explains its power, “Basically, you wouldn’t be able to see from the output of the algorithm if you switched one person’s data with another in your dataset.”
Generalization, she says, is a closely-related concept, “the property of learning algorithms that says the performance of your algorithm on your training set should be close to its performance on unknown data.” explains Zakynthinou “Adaptive processes are common practices in scientific research which reuse datasets for multiple adaptive analyses. They tend to learn too many things about the specific data they’re trained on, which is harmful for both generalization and privacy. Coming up with algorithms that are robust with respect to both Differential Privacy and Generalization is an exciting research goal.”
Kiffer remarks on the fellowship’s value to her research career: “It’s nice that it does give me a little bit of liberty — if I want to be a visiting student at another university, it’s a little bit easier with outside funding.” Since Kiffer is considering post-doc work after graduating, this is especially important to her.
Both Kiffer and Zakynthinou are contemplating careers in research.
Drawn to what she calls academic-style research, Kiffer adds, “I like the open-endedness of research, being able to deep-dive into a single subject. Whatever angle you take with it, it always ends up going somewhere that’s not where you thought it would go.”
And Zakynthinou? She says, “I want to continue towards my current research goals. There are always going to be problems in this field that will be interesting to me. While I am curious to see the problems that people work on in industry, I love theory and want to keep doing that.”
Recently, when the 36 winners of the 2020 Facebook Fellows were announced, and Lucianna “Lulu” Kiffer and Lydia Zakynthinou learned that they each had been awarded one of the prestigious fellowship, they were thrilled. Says Zakynthinou, “It was more than double the happiness.” Both are Khoury College PhD students and colleagues.
While they work in the same lab space on ISEC’s sixth floor, Kiffer and Zakynthinou’s research interests differ dramatically: Kiffer’s work focuses on blockchain protocols, and Zakynthinou’s research, more theory-based, focuses on machine learning and data privacy.
Both students learned of the Facebook Fellowship program, which makes awards to domestic and international students, through an email to Ph.D. students sent by Associate Dean and Professor Frank Tip listing upcoming fellowship and grant opportunities.
The fellowship award includes two years of tuition and fee coverage, an annual stipend, reimbursement for travel expenses, and funding to attend a summit for all 36 members of the fellowship cohort. “Facebook is very into the social part of the fellowship as well as the academic side,” Zakynthinou explains. “They encourage the fellows to reach out to each other and get to know each other.” She has already heard from a handful of other 2020 Facebook Fellows from different universities and research fields across the U.S. and abroad.
The purpose of the fellowship, according to the Facebook Research website, is to “encourage and support promising doctoral students who are engaged in innovative and relevant research.” Fellows have wide-ranging research interests that they are free to continue to pursue for the duration of the fellowship.
Kiffer’s current research on blockchain and cryptocurrency began during early months of her Ph.D. “One of my advisors, Alan Mislove, asked me to read up on this thing, Ethereum, and tell him how it works,” she recalls. “Ethereum is a cryptocurrency. I’d heard about Bitcoin before but never really read much about cryptocurrencies, but I just never stopped. It became my whole Ph.D.”
She believes that the potential of blockchain technology is far-reaching.
“Blockchain is still in its infancy, and there’s a lot to be improved,” she said. “ Even as a currency, Bitcoin has gone where nobody thought it was going to go. There are researchers looking at how to use Bitcoin or similar networks to do more than just currency — things like streaming services with micropayments on peer-to-peer networks, or distributed data storage on personal/work computers. The ability to have a system like this that anybody in the world can opt into is really powerful.”
Zakynthinou, on the other hand, is taking a more theory-based approach to her research, focusing mainly on Differential Privacy and Generalization. Both, she explains, are key aspects of machine learning theory.
“Differential Privacy is a mathematically rigorous technique that has become the standard for ensuring privacy in machine learning models,” Zakynthinou says. “It is based on the idea that if you change one data point in your sample set, then the distribution of the output of the algorithm should not change too much.” She explains its power, “Basically, you wouldn’t be able to see from the output of the algorithm if you switched one person’s data with another in your dataset.”
Generalization, she says, is a closely-related concept, “the property of learning algorithms that says the performance of your algorithm on your training set should be close to its performance on unknown data.” explains Zakynthinou “Adaptive processes are common practices in scientific research which reuse datasets for multiple adaptive analyses. They tend to learn too many things about the specific data they’re trained on, which is harmful for both generalization and privacy. Coming up with algorithms that are robust with respect to both Differential Privacy and Generalization is an exciting research goal.”
Kiffer remarks on the fellowship’s value to her research career: “It’s nice that it does give me a little bit of liberty — if I want to be a visiting student at another university, it’s a little bit easier with outside funding.” Since Kiffer is considering post-doc work after graduating, this is especially important to her.
Both Kiffer and Zakynthinou are contemplating careers in research.
Drawn to what she calls academic-style research, Kiffer adds, “I like the open-endedness of research, being able to deep-dive into a single subject. Whatever angle you take with it, it always ends up going somewhere that’s not where you thought it would go.”
And Zakynthinou? She says, “I want to continue towards my current research goals. There are always going to be problems in this field that will be interesting to me. While I am curious to see the problems that people work on in industry, I love theory and want to keep doing that.”
Recently, when the 36 winners of the 2020 Facebook Fellows were announced, and Lucianna “Lulu” Kiffer and Lydia Zakynthinou learned that they each had been awarded one of the prestigious fellowship, they were thrilled. Says Zakynthinou, “It was more than double the happiness.” Both are Khoury College PhD students and colleagues.
While they work in the same lab space on ISEC’s sixth floor, Kiffer and Zakynthinou’s research interests differ dramatically: Kiffer’s work focuses on blockchain protocols, and Zakynthinou’s research, more theory-based, focuses on machine learning and data privacy.
Both students learned of the Facebook Fellowship program, which makes awards to domestic and international students, through an email to Ph.D. students sent by Associate Dean and Professor Frank Tip listing upcoming fellowship and grant opportunities.
The fellowship award includes two years of tuition and fee coverage, an annual stipend, reimbursement for travel expenses, and funding to attend a summit for all 36 members of the fellowship cohort. “Facebook is very into the social part of the fellowship as well as the academic side,” Zakynthinou explains. “They encourage the fellows to reach out to each other and get to know each other.” She has already heard from a handful of other 2020 Facebook Fellows from different universities and research fields across the U.S. and abroad.
The purpose of the fellowship, according to the Facebook Research website, is to “encourage and support promising doctoral students who are engaged in innovative and relevant research.” Fellows have wide-ranging research interests that they are free to continue to pursue for the duration of the fellowship.
Kiffer’s current research on blockchain and cryptocurrency began during early months of her Ph.D. “One of my advisors, Alan Mislove, asked me to read up on this thing, Ethereum, and tell him how it works,” she recalls. “Ethereum is a cryptocurrency. I’d heard about Bitcoin before but never really read much about cryptocurrencies, but I just never stopped. It became my whole Ph.D.”
She believes that the potential of blockchain technology is far-reaching.
“Blockchain is still in its infancy, and there’s a lot to be improved,” she said. “ Even as a currency, Bitcoin has gone where nobody thought it was going to go. There are researchers looking at how to use Bitcoin or similar networks to do more than just currency — things like streaming services with micropayments on peer-to-peer networks, or distributed data storage on personal/work computers. The ability to have a system like this that anybody in the world can opt into is really powerful.”
Zakynthinou, on the other hand, is taking a more theory-based approach to her research, focusing mainly on Differential Privacy and Generalization. Both, she explains, are key aspects of machine learning theory.
“Differential Privacy is a mathematically rigorous technique that has become the standard for ensuring privacy in machine learning models,” Zakynthinou says. “It is based on the idea that if you change one data point in your sample set, then the distribution of the output of the algorithm should not change too much.” She explains its power, “Basically, you wouldn’t be able to see from the output of the algorithm if you switched one person’s data with another in your dataset.”
Generalization, she says, is a closely-related concept, “the property of learning algorithms that says the performance of your algorithm on your training set should be close to its performance on unknown data.” explains Zakynthinou “Adaptive processes are common practices in scientific research which reuse datasets for multiple adaptive analyses. They tend to learn too many things about the specific data they’re trained on, which is harmful for both generalization and privacy. Coming up with algorithms that are robust with respect to both Differential Privacy and Generalization is an exciting research goal.”
Kiffer remarks on the fellowship’s value to her research career: “It’s nice that it does give me a little bit of liberty — if I want to be a visiting student at another university, it’s a little bit easier with outside funding.” Since Kiffer is considering post-doc work after graduating, this is especially important to her.
Both Kiffer and Zakynthinou are contemplating careers in research.
Drawn to what she calls academic-style research, Kiffer adds, “I like the open-endedness of research, being able to deep-dive into a single subject. Whatever angle you take with it, it always ends up going somewhere that’s not where you thought it would go.”
And Zakynthinou? She says, “I want to continue towards my current research goals. There are always going to be problems in this field that will be interesting to me. While I am curious to see the problems that people work on in industry, I love theory and want to keep doing that.”
Recently, when the 36 winners of the 2020 Facebook Fellows were announced, and Lucianna “Lulu” Kiffer and Lydia Zakynthinou learned that they each had been awarded one of the prestigious fellowship, they were thrilled. Says Zakynthinou, “It was more than double the happiness.” Both are Khoury College PhD students and colleagues.
While they work in the same lab space on ISEC’s sixth floor, Kiffer and Zakynthinou’s research interests differ dramatically: Kiffer’s work focuses on blockchain protocols, and Zakynthinou’s research, more theory-based, focuses on machine learning and data privacy.
Both students learned of the Facebook Fellowship program, which makes awards to domestic and international students, through an email to Ph.D. students sent by Associate Dean and Professor Frank Tip listing upcoming fellowship and grant opportunities.
The fellowship award includes two years of tuition and fee coverage, an annual stipend, reimbursement for travel expenses, and funding to attend a summit for all 36 members of the fellowship cohort. “Facebook is very into the social part of the fellowship as well as the academic side,” Zakynthinou explains. “They encourage the fellows to reach out to each other and get to know each other.” She has already heard from a handful of other 2020 Facebook Fellows from different universities and research fields across the U.S. and abroad.
The purpose of the fellowship, according to the Facebook Research website, is to “encourage and support promising doctoral students who are engaged in innovative and relevant research.” Fellows have wide-ranging research interests that they are free to continue to pursue for the duration of the fellowship.
Kiffer’s current research on blockchain and cryptocurrency began during early months of her Ph.D. “One of my advisors, Alan Mislove, asked me to read up on this thing, Ethereum, and tell him how it works,” she recalls. “Ethereum is a cryptocurrency. I’d heard about Bitcoin before but never really read much about cryptocurrencies, but I just never stopped. It became my whole Ph.D.”
She believes that the potential of blockchain technology is far-reaching.
“Blockchain is still in its infancy, and there’s a lot to be improved,” she said. “ Even as a currency, Bitcoin has gone where nobody thought it was going to go. There are researchers looking at how to use Bitcoin or similar networks to do more than just currency — things like streaming services with micropayments on peer-to-peer networks, or distributed data storage on personal/work computers. The ability to have a system like this that anybody in the world can opt into is really powerful.”
Zakynthinou, on the other hand, is taking a more theory-based approach to her research, focusing mainly on Differential Privacy and Generalization. Both, she explains, are key aspects of machine learning theory.
“Differential Privacy is a mathematically rigorous technique that has become the standard for ensuring privacy in machine learning models,” Zakynthinou says. “It is based on the idea that if you change one data point in your sample set, then the distribution of the output of the algorithm should not change too much.” She explains its power, “Basically, you wouldn’t be able to see from the output of the algorithm if you switched one person’s data with another in your dataset.”
Generalization, she says, is a closely-related concept, “the property of learning algorithms that says the performance of your algorithm on your training set should be close to its performance on unknown data.” explains Zakynthinou “Adaptive processes are common practices in scientific research which reuse datasets for multiple adaptive analyses. They tend to learn too many things about the specific data they’re trained on, which is harmful for both generalization and privacy. Coming up with algorithms that are robust with respect to both Differential Privacy and Generalization is an exciting research goal.”
Kiffer remarks on the fellowship’s value to her research career: “It’s nice that it does give me a little bit of liberty — if I want to be a visiting student at another university, it’s a little bit easier with outside funding.” Since Kiffer is considering post-doc work after graduating, this is especially important to her.
Both Kiffer and Zakynthinou are contemplating careers in research.
Drawn to what she calls academic-style research, Kiffer adds, “I like the open-endedness of research, being able to deep-dive into a single subject. Whatever angle you take with it, it always ends up going somewhere that’s not where you thought it would go.”
And Zakynthinou? She says, “I want to continue towards my current research goals. There are always going to be problems in this field that will be interesting to me. While I am curious to see the problems that people work on in industry, I love theory and want to keep doing that.”
Recently, when the 36 winners of the 2020 Facebook Fellows were announced, and Lucianna “Lulu” Kiffer and Lydia Zakynthinou learned that they each had been awarded one of the prestigious fellowship, they were thrilled. Says Zakynthinou, “It was more than double the happiness.” Both are Khoury College PhD students and colleagues.
While they work in the same lab space on ISEC’s sixth floor, Kiffer and Zakynthinou’s research interests differ dramatically: Kiffer’s work focuses on blockchain protocols, and Zakynthinou’s research, more theory-based, focuses on machine learning and data privacy.
Both students learned of the Facebook Fellowship program, which makes awards to domestic and international students, through an email to Ph.D. students sent by Associate Dean and Professor Frank Tip listing upcoming fellowship and grant opportunities.
The fellowship award includes two years of tuition and fee coverage, an annual stipend, reimbursement for travel expenses, and funding to attend a summit for all 36 members of the fellowship cohort. “Facebook is very into the social part of the fellowship as well as the academic side,” Zakynthinou explains. “They encourage the fellows to reach out to each other and get to know each other.” She has already heard from a handful of other 2020 Facebook Fellows from different universities and research fields across the U.S. and abroad.
The purpose of the fellowship, according to the Facebook Research website, is to “encourage and support promising doctoral students who are engaged in innovative and relevant research.” Fellows have wide-ranging research interests that they are free to continue to pursue for the duration of the fellowship.
Kiffer’s current research on blockchain and cryptocurrency began during early months of her Ph.D. “One of my advisors, Alan Mislove, asked me to read up on this thing, Ethereum, and tell him how it works,” she recalls. “Ethereum is a cryptocurrency. I’d heard about Bitcoin before but never really read much about cryptocurrencies, but I just never stopped. It became my whole Ph.D.”
She believes that the potential of blockchain technology is far-reaching.
“Blockchain is still in its infancy, and there’s a lot to be improved,” she said. “ Even as a currency, Bitcoin has gone where nobody thought it was going to go. There are researchers looking at how to use Bitcoin or similar networks to do more than just currency — things like streaming services with micropayments on peer-to-peer networks, or distributed data storage on personal/work computers. The ability to have a system like this that anybody in the world can opt into is really powerful.”
Zakynthinou, on the other hand, is taking a more theory-based approach to her research, focusing mainly on Differential Privacy and Generalization. Both, she explains, are key aspects of machine learning theory.
“Differential Privacy is a mathematically rigorous technique that has become the standard for ensuring privacy in machine learning models,” Zakynthinou says. “It is based on the idea that if you change one data point in your sample set, then the distribution of the output of the algorithm should not change too much.” She explains its power, “Basically, you wouldn’t be able to see from the output of the algorithm if you switched one person’s data with another in your dataset.”
Generalization, she says, is a closely-related concept, “the property of learning algorithms that says the performance of your algorithm on your training set should be close to its performance on unknown data.” explains Zakynthinou “Adaptive processes are common practices in scientific research which reuse datasets for multiple adaptive analyses. They tend to learn too many things about the specific data they’re trained on, which is harmful for both generalization and privacy. Coming up with algorithms that are robust with respect to both Differential Privacy and Generalization is an exciting research goal.”
Kiffer remarks on the fellowship’s value to her research career: “It’s nice that it does give me a little bit of liberty — if I want to be a visiting student at another university, it’s a little bit easier with outside funding.” Since Kiffer is considering post-doc work after graduating, this is especially important to her.
Both Kiffer and Zakynthinou are contemplating careers in research.
Drawn to what she calls academic-style research, Kiffer adds, “I like the open-endedness of research, being able to deep-dive into a single subject. Whatever angle you take with it, it always ends up going somewhere that’s not where you thought it would go.”
And Zakynthinou? She says, “I want to continue towards my current research goals. There are always going to be problems in this field that will be interesting to me. While I am curious to see the problems that people work on in industry, I love theory and want to keep doing that.”
Recently, when the 36 winners of the 2020 Facebook Fellows were announced, and Lucianna “Lulu” Kiffer and Lydia Zakynthinou learned that they each had been awarded one of the prestigious fellowship, they were thrilled. Says Zakynthinou, “It was more than double the happiness.” Both are Khoury College PhD students and colleagues.
While they work in the same lab space on ISEC’s sixth floor, Kiffer and Zakynthinou’s research interests differ dramatically: Kiffer’s work focuses on blockchain protocols, and Zakynthinou’s research, more theory-based, focuses on machine learning and data privacy.
Both students learned of the Facebook Fellowship program, which makes awards to domestic and international students, through an email to Ph.D. students sent by Associate Dean and Professor Frank Tip listing upcoming fellowship and grant opportunities.
The fellowship award includes two years of tuition and fee coverage, an annual stipend, reimbursement for travel expenses, and funding to attend a summit for all 36 members of the fellowship cohort. “Facebook is very into the social part of the fellowship as well as the academic side,” Zakynthinou explains. “They encourage the fellows to reach out to each other and get to know each other.” She has already heard from a handful of other 2020 Facebook Fellows from different universities and research fields across the U.S. and abroad.
The purpose of the fellowship, according to the Facebook Research website, is to “encourage and support promising doctoral students who are engaged in innovative and relevant research.” Fellows have wide-ranging research interests that they are free to continue to pursue for the duration of the fellowship.
Kiffer’s current research on blockchain and cryptocurrency began during early months of her Ph.D. “One of my advisors, Alan Mislove, asked me to read up on this thing, Ethereum, and tell him how it works,” she recalls. “Ethereum is a cryptocurrency. I’d heard about Bitcoin before but never really read much about cryptocurrencies, but I just never stopped. It became my whole Ph.D.”
She believes that the potential of blockchain technology is far-reaching.
“Blockchain is still in its infancy, and there’s a lot to be improved,” she said. “ Even as a currency, Bitcoin has gone where nobody thought it was going to go. There are researchers looking at how to use Bitcoin or similar networks to do more than just currency — things like streaming services with micropayments on peer-to-peer networks, or distributed data storage on personal/work computers. The ability to have a system like this that anybody in the world can opt into is really powerful.”
Zakynthinou, on the other hand, is taking a more theory-based approach to her research, focusing mainly on Differential Privacy and Generalization. Both, she explains, are key aspects of machine learning theory.
“Differential Privacy is a mathematically rigorous technique that has become the standard for ensuring privacy in machine learning models,” Zakynthinou says. “It is based on the idea that if you change one data point in your sample set, then the distribution of the output of the algorithm should not change too much.” She explains its power, “Basically, you wouldn’t be able to see from the output of the algorithm if you switched one person’s data with another in your dataset.”
Generalization, she says, is a closely-related concept, “the property of learning algorithms that says the performance of your algorithm on your training set should be close to its performance on unknown data.” explains Zakynthinou “Adaptive processes are common practices in scientific research which reuse datasets for multiple adaptive analyses. They tend to learn too many things about the specific data they’re trained on, which is harmful for both generalization and privacy. Coming up with algorithms that are robust with respect to both Differential Privacy and Generalization is an exciting research goal.”
Kiffer remarks on the fellowship’s value to her research career: “It’s nice that it does give me a little bit of liberty — if I want to be a visiting student at another university, it’s a little bit easier with outside funding.” Since Kiffer is considering post-doc work after graduating, this is especially important to her.
Both Kiffer and Zakynthinou are contemplating careers in research.
Drawn to what she calls academic-style research, Kiffer adds, “I like the open-endedness of research, being able to deep-dive into a single subject. Whatever angle you take with it, it always ends up going somewhere that’s not where you thought it would go.”
And Zakynthinou? She says, “I want to continue towards my current research goals. There are always going to be problems in this field that will be interesting to me. While I am curious to see the problems that people work on in industry, I love theory and want to keep doing that.”
Recently, when the 36 winners of the 2020 Facebook Fellows were announced, and Lucianna “Lulu” Kiffer and Lydia Zakynthinou learned that they each had been awarded one of the prestigious fellowship, they were thrilled. Says Zakynthinou, “It was more than double the happiness.” Both are Khoury College PhD students and colleagues.
While they work in the same lab space on ISEC’s sixth floor, Kiffer and Zakynthinou’s research interests differ dramatically: Kiffer’s work focuses on blockchain protocols, and Zakynthinou’s research, more theory-based, focuses on machine learning and data privacy.
Both students learned of the Facebook Fellowship program, which makes awards to domestic and international students, through an email to Ph.D. students sent by Associate Dean and Professor Frank Tip listing upcoming fellowship and grant opportunities.
The fellowship award includes two years of tuition and fee coverage, an annual stipend, reimbursement for travel expenses, and funding to attend a summit for all 36 members of the fellowship cohort. “Facebook is very into the social part of the fellowship as well as the academic side,” Zakynthinou explains. “They encourage the fellows to reach out to each other and get to know each other.” She has already heard from a handful of other 2020 Facebook Fellows from different universities and research fields across the U.S. and abroad.
The purpose of the fellowship, according to the Facebook Research website, is to “encourage and support promising doctoral students who are engaged in innovative and relevant research.” Fellows have wide-ranging research interests that they are free to continue to pursue for the duration of the fellowship.
Kiffer’s current research on blockchain and cryptocurrency began during early months of her Ph.D. “One of my advisors, Alan Mislove, asked me to read up on this thing, Ethereum, and tell him how it works,” she recalls. “Ethereum is a cryptocurrency. I’d heard about Bitcoin before but never really read much about cryptocurrencies, but I just never stopped. It became my whole Ph.D.”
She believes that the potential of blockchain technology is far-reaching.
“Blockchain is still in its infancy, and there’s a lot to be improved,” she said. “ Even as a currency, Bitcoin has gone where nobody thought it was going to go. There are researchers looking at how to use Bitcoin or similar networks to do more than just currency — things like streaming services with micropayments on peer-to-peer networks, or distributed data storage on personal/work computers. The ability to have a system like this that anybody in the world can opt into is really powerful.”
Zakynthinou, on the other hand, is taking a more theory-based approach to her research, focusing mainly on Differential Privacy and Generalization. Both, she explains, are key aspects of machine learning theory.
“Differential Privacy is a mathematically rigorous technique that has become the standard for ensuring privacy in machine learning models,” Zakynthinou says. “It is based on the idea that if you change one data point in your sample set, then the distribution of the output of the algorithm should not change too much.” She explains its power, “Basically, you wouldn’t be able to see from the output of the algorithm if you switched one person’s data with another in your dataset.”
Generalization, she says, is a closely-related concept, “the property of learning algorithms that says the performance of your algorithm on your training set should be close to its performance on unknown data.” explains Zakynthinou “Adaptive processes are common practices in scientific research which reuse datasets for multiple adaptive analyses. They tend to learn too many things about the specific data they’re trained on, which is harmful for both generalization and privacy. Coming up with algorithms that are robust with respect to both Differential Privacy and Generalization is an exciting research goal.”
Kiffer remarks on the fellowship’s value to her research career: “It’s nice that it does give me a little bit of liberty — if I want to be a visiting student at another university, it’s a little bit easier with outside funding.” Since Kiffer is considering post-doc work after graduating, this is especially important to her.
Both Kiffer and Zakynthinou are contemplating careers in research.
Drawn to what she calls academic-style research, Kiffer adds, “I like the open-endedness of research, being able to deep-dive into a single subject. Whatever angle you take with it, it always ends up going somewhere that’s not where you thought it would go.”
And Zakynthinou? She says, “I want to continue towards my current research goals. There are always going to be problems in this field that will be interesting to me. While I am curious to see the problems that people work on in industry, I love theory and want to keep doing that.”
Recently, when the 36 winners of the 2020 Facebook Fellows were announced, and Lucianna “Lulu” Kiffer and Lydia Zakynthinou learned that they each had been awarded one of the prestigious fellowship, they were thrilled. Says Zakynthinou, “It was more than double the happiness.” Both are Khoury College PhD students and colleagues.
While they work in the same lab space on ISEC’s sixth floor, Kiffer and Zakynthinou’s research interests differ dramatically: Kiffer’s work focuses on blockchain protocols, and Zakynthinou’s research, more theory-based, focuses on machine learning and data privacy.
Both students learned of the Facebook Fellowship program, which makes awards to domestic and international students, through an email to Ph.D. students sent by Associate Dean and Professor Frank Tip listing upcoming fellowship and grant opportunities.
The fellowship award includes two years of tuition and fee coverage, an annual stipend, reimbursement for travel expenses, and funding to attend a summit for all 36 members of the fellowship cohort. “Facebook is very into the social part of the fellowship as well as the academic side,” Zakynthinou explains. “They encourage the fellows to reach out to each other and get to know each other.” She has already heard from a handful of other 2020 Facebook Fellows from different universities and research fields across the U.S. and abroad.
The purpose of the fellowship, according to the Facebook Research website, is to “encourage and support promising doctoral students who are engaged in innovative and relevant research.” Fellows have wide-ranging research interests that they are free to continue to pursue for the duration of the fellowship.
Kiffer’s current research on blockchain and cryptocurrency began during early months of her Ph.D. “One of my advisors, Alan Mislove, asked me to read up on this thing, Ethereum, and tell him how it works,” she recalls. “Ethereum is a cryptocurrency. I’d heard about Bitcoin before but never really read much about cryptocurrencies, but I just never stopped. It became my whole Ph.D.”
She believes that the potential of blockchain technology is far-reaching.
“Blockchain is still in its infancy, and there’s a lot to be improved,” she said. “ Even as a currency, Bitcoin has gone where nobody thought it was going to go. There are researchers looking at how to use Bitcoin or similar networks to do more than just currency — things like streaming services with micropayments on peer-to-peer networks, or distributed data storage on personal/work computers. The ability to have a system like this that anybody in the world can opt into is really powerful.”
Zakynthinou, on the other hand, is taking a more theory-based approach to her research, focusing mainly on Differential Privacy and Generalization. Both, she explains, are key aspects of machine learning theory.
“Differential Privacy is a mathematically rigorous technique that has become the standard for ensuring privacy in machine learning models,” Zakynthinou says. “It is based on the idea that if you change one data point in your sample set, then the distribution of the output of the algorithm should not change too much.” She explains its power, “Basically, you wouldn’t be able to see from the output of the algorithm if you switched one person’s data with another in your dataset.”
Generalization, she says, is a closely-related concept, “the property of learning algorithms that says the performance of your algorithm on your training set should be close to its performance on unknown data.” explains Zakynthinou “Adaptive processes are common practices in scientific research which reuse datasets for multiple adaptive analyses. They tend to learn too many things about the specific data they’re trained on, which is harmful for both generalization and privacy. Coming up with algorithms that are robust with respect to both Differential Privacy and Generalization is an exciting research goal.”
Kiffer remarks on the fellowship’s value to her research career: “It’s nice that it does give me a little bit of liberty — if I want to be a visiting student at another university, it’s a little bit easier with outside funding.” Since Kiffer is considering post-doc work after graduating, this is especially important to her.
Both Kiffer and Zakynthinou are contemplating careers in research.
Drawn to what she calls academic-style research, Kiffer adds, “I like the open-endedness of research, being able to deep-dive into a single subject. Whatever angle you take with it, it always ends up going somewhere that’s not where you thought it would go.”
And Zakynthinou? She says, “I want to continue towards my current research goals. There are always going to be problems in this field that will be interesting to me. While I am curious to see the problems that people work on in industry, I love theory and want to keep doing that.”
Khoury Social
Contact Khoury
Khoury Social
Contact Khoury
Khoury Social
Contact Khoury
Khoury Social
Contact Khoury