Christopher Kanan, Ph.D. - US grants

This proposal requests funds to support the annual Inter-Science of Learning Centers (i-SLC) Conference, organized by the graduate students and postdoctoral fellows of the 6 Science of Learning Centers (SLCs). This event has served as a vital venue for exchanging research findings across the SLCs, sharing resources and stimulating career development opportunities for Center trainees. This highly productive activity has resulted in cross-center collaborations, including cross-center student exchanges for additional training in other laboratories.

This year, the conference will be hosted by the NSF-funded Temporal Dynamics of Learning Center (TDLC) in San Diego, CA on April 21-23, 2012. To capitalize on the host-center's primary research focus on the temporal dynamics of learning and how time and timing influences learning, the iSLC 2012 will have the general theme of "Time, Mind, and Education Interwined". Requested funds will be used to pay for conference facilities at the University of California San Diego (UCSD) and to cover participants' costs of attending, including travel and per diem

Earlier iSLC conferences experiment in various ways to maximize the effectiveness with which the meeting meets its objectives, and each evolution of this event reflects deliberate efforts to improve the meeting based on feedback from prior meetings. Individual students and postdoctoral fellows attending iSLC will benefit from visiting UCSD, learning from experiences of their peers, and by being exposed to new and alternative methodologies and paradigms. Most importantly, they will be provided with opportunities to participate actively in the cross-center, and inter-disciplinary network that will provide support for them in their careers. Since most graduate students and postdoctoral fellows typically attend field-specific conferences with little opportunity to interact with peers outside of their disciplinary niches, this annual event is addressing the need to provide the infrastructure for a national, and even international, interdisciplinary network of young scholars who are likely to continue to cooperate and collaborate throughout their future careers. This year, the trainees are proactively implementing several strategies to broaden the participation of underrepresented minorities in science, by leveraging on-going efforts at their respective centers.

While machine learning and artificial intelligence has greatly advanced in recent years, these systems still have significant limitations. Machine learning systems have distinct learning and deployment phases. If new information is acquired, the entire system is often rebuilt rather than having only the new information being learned because otherwise the system will forget a large amount of its past knowledge. Systems cannot learn autonomously and often require strong supervision. This project aims to address these issues by creating new multi-modal brain-inspired algorithms capable of learning immediately without excess forgetting. These algorithms can enable learning with fewer computational resources, which can facilitate learning on devices such as cell phones and home robots. Fast learning from multimodal data streams is critical to enabling natural interactions with artificial agents. Autonomous multimodal learning will reduce reliance on annotated data, which is a huge bottleneck in increasing the utility of artificial intelligence, and may enable significant gains in performance. This research will provide building blocks that others can use to create new algorithms, applications, and cognitive technologies.

The algorithms are based on the complementary learning systems theory for how the human brain learns quickly. The human brain uses its hippocampus to immediately learn new information and then this information is transferred to the neocortex during sleep. Based on this theory, streaming learning algorithms for deep neural networks will be created, which will enable fast learning from structured data streams without catastrophic forgetting of past knowledge. The algorithms will be assessed based on their ability to classify large image databases containing thousands of categories. These systems will be leveraged to pioneer multimodal streaming learning for visual question answering and visual query detection, enabling language to inform understanding of visual scenes. These traits will be integrated to enable a model to autonomously query an environment with limited human supervision.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.