DysDiTect: Dyslexia Identification Using CNN-Positional-LSTM-Attention Modeling with Chinese Dictation Task
Chinese dyslexic children often write with frequent errors, including reversed characters and substitution of radicals. Then, is it possible to use deep learning model to identify the patterns exhibited by these children? Our lab’s machine learning and dyslexia work by graduate student Hey Wing Liu demonstrated that our new model, DysDiTect, can effectively identify children with dyslexia considering the sequential temporal properties in a Chinese dictation task. Our results revealed the possibilities to discover intrinsic properties in dyslexic writings, and enabled fast and accurate predictions of children at risk of dyslexia. This paper was published in Brain Sciences on April 29th, 2024, and can be accessed using the following link: https://doi.org/10.3390/brainsci14050444
Unraveling Temporal Dynamics of Multidimensional Statistical Learning in Implicit and Explicit Systems: An X-Way Hypothesis
Our brains receive different kinds of information every second, where patterns exist. Have you wondered how we efficiently process tremendous information from the environment and adapt to the patterns? Using a novel multidimensional statistical learning task, the work of Dr. Stephen Man Kit Lee, a postdoctoral fellow in our SLR lab, demonstrated that humans can simultaneously acquire multiple regularities in different stages.
The experimental findings were integrated into a new hypothesis explaining the mechanisms underlying multidimensional statistical learning. This paper was published in Cognitive Science on April 2nd, 2024, and can be accessed using the following link: http://dx.doi.org/10.1111/cogs.13437
Cue Predictiveness and Uncertainty Determine Cue Representation During Visual Statistical Learning
Our SLR lab’s PhD student, Puyuan's work was featured on the cover page of Learning & Memory. Please check out Puyuan's paper titled "Cue Predictiveness and Uncertainty Determine Cue Representation During Visual Statistical Learning"!
In this paper, Puyuan developed a novel probabilistic cueing-validation paradigm and demonstrated that input uncertainty regulates the operation of exploration-like (relatively implicit) and exploitation-like (relatively explicit) cue processing during visual statistical learning.
Thanks to Mei Zhou, Arpitha Vasudevamurthy, and Stephen Man Kit Lee for the great illustration of the cover page! This paper was published in Learning & Memory in 2023, and can be accessed using the following link: http://www.learnmem.org/cgi/doi/10.1101/lm.053777.123
