1/30/2023 Aaron Seidlitz, Illinois CS 5 min read
The University of Illinois Urbana-Champaign team, CharmBana, is led by Illinois CS Ph.D. student Revanth Gangi Reddi and includes twelve other students from CS, Electrical & Computer Engineering, and the iSchool. It is coached by CS professor ChengXiang Zhai.
Written by Aaron Seidlitz, Illinois CS
Now in his first year as a Ph.D. student, Revanth Gangi Reddy recalled that he chose Illinois Computer Science because of the reputation built here for strong research within Artificial Intelligence, especially regarding his focus on Natural Language Processing (NLP).
Beginning with his undergraduate studies at the Indian Institute of Technology, Madras and continuing through his AI Resident role at the IBM Research New York, Revanth has centered his research into two subfields of NLP – Question Answering (QA) and Information Retrieval (IR).
As time has gone on, he has even considered developing a social chatbot a “pet-project” of his.
Now, that pet-project has become a reality, as Revanth leads the CharmBana team in the Alexa Prize SocialBot Grant Challenge 5.
The 13-student team will be coached by Illinois CS professor ChengXiang Zhai, and its students span the CS and Electrical & Computer Engineering departments within The Grainger College of Engineering, as well as the School of Information Sciences.
“The competition is a great opportunity for us to make high-visibility impact in the field of conversational AI,” said Revanth, whose Ph.D advisor is professor Heng Ji. “This experience provides a unique advantage of being able to obtain feedback from Amazon Alexa’s wide set of users. Such real-life feedback is usually very hard to obtain in any other settings, hence we intend to fully incorporate end-user feedback into a lot of the design and modeling decisions that we make along the course of the competition.
“This competition also provides us with mentorship from Amazon Alexa scientists, and we’re really excited to brainstorm and bounce ideas off of them.”
Amazon announced the participants toward the end of December 2022. Nine teams from around the globe have been selected to participate in the university challenge focused on advancing human-computer interaction and open dialogue conversation.
Alexa Prize is a flagship industry-academic partnership program dedicated to accelerating the science of conversational AI and multimodal human-AI interactions.
The teams selected for the challenge include five returning competitors — including the top two finishers in the most recent challenge — and four new universities.
The Alexa Prize, launched in 2016, challenges university teams to design socialbots that Amazon customers can interact with via Alexa-enabled devices.
The ultimate goal is to meet the Grand Challenge: earn a composite score of 4.0 or higher (out of 5) from a panel of judges, and have those judges find that at least two-thirds of their conversations with the socialbot in the final round of judging remain coherent and engaging for 20 minutes.
The first team to meet the Grand Challenge will win a $1 million research grant for their university.
“Prize competitions provide an agile experimentation framework for researchers and students, encouraging them to explore transformational ideas at the boundaries of what is achievable,” said Reza Ghanadan, senior principal scientist with Amazon Alexa AI and head of Alexa Prize.
This year there are two sets of awards: one set for overall performance and one set for scientific impact. Prizes for overall performance in the competition will be $250,000 for the first-place team, $50,000 for second, and $25,000 for third.
In an effort to sharpen the focus on scientific contribution and advance the science behind conversational AI, this year’s university teams will not only compete for prizes related to overall performance, but also for a separate award for scientific invention and innovation.
The new award allows teams to focus on advancing the field of conversational AI through a deeper study of the fundamentals of open dialogue conversations. Cash prizes for scientific contribution will be awarded to students on the winning teams — $250,000 for the first-place team, $50,000 for second, and $25,000 for third.
Each university selected for the challenge receives a research grant of up to $250,000, Alexa-enabled devices, free Amazon Web Services (AWS) cloud computing services to support their research and development efforts, access to Amazon scientists, the CoBot (conversational bot) toolkit, and other tools such as automated speech recognition through Alexa, neural detection and generation models, conversational data sets, and design guidance and development support from the Alexa Prize team.
In previous challenges, participating teams have improved the state of the art for open domain dialogue systems by developing improved natural language understanding (NLU) systems, neural response generation models, common sense knowledge modeling, and dialogue policies leading to smoother, and more engaging conversations.
The teams’ socialbots will be available for Alexa customers to engage with in March 2023, a finals competition will be held in early August 2023, and winners announced later that month.
“As a new entrant into this competition, I feel that we have our task cut out to make the finals. However, starting off with a clean slate can help with innovation, and we’re looking forward to realizing some of the cool ideas that we put forward in the proposal,” Revanth said. “The team has a good mix of experience, enthusiasm and energy from grad students and undergrads. We started off with a core team initially and expanded along the way as we identified new avenues that we could explore.”
The “Alquist” team from Czech Technical University won the fourth challenge, with teams from Stanford and the University of Buffalo earning second- and third-place prizes, respectively. The publications from that challenge can be found here.
Winning teams from previous years include Emory University, the University of Washington, and the University of California, Davis.