My goal is to boost human learning and
performance by developing and applying formal models of
cognition. I am interested in producing software that
enables individuals to learn and perform tasks efficiently
and effortlessly. My approach draws on methods from machine
learning and theories from cognitive science in order to
construct robust psychological models that characterize
the computational challenges faced by an individual
attempting to complete a task. My research lies at the
interface of human learning, machine learning, and
computer-assisted decision making.
My research has predominantly focused on
helping individuals categorize visual images. I have
approached this objective from two perspectives: decision
support and efficient training. Decision support enables
expert-like levels of performance—without training—by
exploiting ordinary but powerful human visual capabilities.
Efficient training promotes the discovery of the visual
features necessary to correctly categorize the images.
Both approaches leverage a latent space representation of
human-perceived similarity, which we refer to as a
Current limitations in visual task performance motivate my primary research questions:
- What kinds of visual tasks would benefit most from a computer-assisted training paradigm?
- What types of training will make acquiring visual expertise more efficient?
- What types of decisional support will enable relatively effortless but accurate performance on visual tasks?
Complete Research Statement [ PDF ]
Current CV [ PDF ]
- Received a joint Ph.D. in Computer Science and Cognitive Science from CU Boulder (2017)
- Received a M.S. in Computer Science from CU Boulder (2013)
- Received a B.S. in Engineering Physics from CU Boulder (2011)
Institute of Cognitive Science at the University of Colorado Boulder
- Roads, B. D., & Mozer, M. C. (in preparation). Using human-surrogate models to optimize training sequences during visual category training.
- Roads, B. D., & Mozer, M. C. (in preparation). Using enriched training environments for visual category training.
- Roads, B. D., & Mozer, M. C. (in preparation). Predicting the Difficulty of Human Category Learning Using Exemplar-Based Neural Networks.
- Roads, B. D., & Mozer, M. C. (in preparation). Obtaining psychological embeddings through joint kernel and metric learning.
- Roads, B. D., Xu, B., Robinson, J. K., & Tanaka, J. W. (submitted). The Easy-to-Hard Training Advantage with Real-World Medical Images.
- Snell, J., Ridgeway, K., Liao, R., Roads, B. D., Mozer, M. C., & Zemel, R. S. (2017). Learning to generate images with perceptual similarity metrics. Accepted for publication in IEEE International Conference on Image Processing. arXiv:1511.06409v3
- Roads, B. D., & Mozer, M. C. (2017). Improving human-machine cooperative classification via cognitive theories of similarity. Cognitive Science: A Multidisciplinary Journal, 41, 1394-1411. DOI: 10.1111/cogs.12400. [ PDF ]
- Khajah, M., Roads, B. D., Lindsey, R. V., Liu, Y.-E., & Mozer, M. C. (2016). Designing engaging games using Bayesian optimization. In Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems (pp. 5571-5582). New York: ACM. [ PDF ]
- Roads, B. D., Mozer, M. C., & Busey, T. A. (2016). Using highlighting to train attentional expertise. PLoS ONE 11(1): e0146266. doi:10.1371/journal.pone.0146266. [ PDF ]
Although my passion for my work is difficult to rigorously justify, it is arguably the most indispensable aspect of my research. My passion is supported by two fundamental pillars: the flexibility of cognition, nature
A hero from a modern day myth said, "Your focus determines your reality."
I believe 'focus' in this perceptual sense is strongly driven my habitual factors.
Unification of strong applications and theoretical frameworks