Insights | DSI

Skip to main content

Sort:

Figures 1a-b show the general strategy: during a training period (fig. 1a), a time-dependent state is imposed on the system. This is pictured by an origami bird whose wings are periodically moved by an external agent. After training (fig. 1b), a retrieval period during which the physical system evolves under the learned dynamics should lead to a dynamic state that matches the training as best as possible. (fig. 1c and 1d) Demonstration of the training and retrieval procedure using a programmable LEGO toy. The angular positions of two motors are imposed by hand during a training phase (fig. 1c), during which couplings between the motors are learned. Dynamics during retrieval, with learned couplings, can produce fixed points as well as dynamic states where the angles constantly change (fig. 1d). See this video for a detailed description.

BlogSep 03, 2024

Developing a smart material that can learn from example

Fig 1. Percentage of broadband-serviceable locations (BSLs) with at least one submitted challenge per US state or territory. Nebraska had strong participation in the challenge process, followed by Virginia, New Mexico, and Maine. Most states and territories have low participation rates.

BlogAug 29, 2024

The Challenging Path to Internet Broadband Access Equity

BlogJul 30, 2024

AI Meets Mechanobiology: Decoding the Mechanics of Living Systems

BlogJul 29, 2024

Does Political Polarization Really Undermine Democracy? Maybe Not

A video frame example from Cityscape dataset where different colors denote different labels in pixel level. For example, red means person and light blue means sky.

BlogJun 06, 2024

Behind Automatic Video Semantic Segmentation

Microbial organisms drive the carbon cycle. Image adapted from Wu et al. 2024 https://doi.org/10.1016/j.scitotenv.2023.168627

BlogMay 30, 2024

Searching for simplicity in microbial communities

Fig: Formation of HIV capsid simulated by a coarse-grained model.
From: https://www.google.com/url?q=https://www.science.org/doi/epdf/10.1126/sciadv.add7434&sa=D&source=docs&ust=1715014663521469&usg=AOvVaw3fATTBC_yAvxPv6WPYI17_

BlogMay 22, 2024

AI as a Great Teacher for Molecular Dynamic Modeling

AI methods and models have enabled a huge leap in our understanding of how images are processed in the brain.
We used to describe visual neurons as “edge detectors” and “face detectors”. Using deep neural networks, we have
discovered that images like these (which really can’t be described with words) are richer models of single neurons in
our visual system. I liken these AI-enabled descriptions of neural function, perhaps ironically, to a whole new kind of
vocabulary that neuroscientists can now use to explain the visual system. (Images from various papers including a,b, c, d, e , f)

BlogMay 16, 2024

Expanding Our Vocabulary of Vision Using AI

Caption: (left) We want to find networks which can easily switch between two different motions. (right)
Alternating the design process for each motion results in networks which can easily switch between the two motions.

BlogMay 09, 2024

Teaching materials to adapt

Fig: The neural network will be trained on a dataset generated by validated quantum mechanics methods on small molecules. The model will then be used to predict the x-ray damage effect in more complex systems.

BlogMay 02, 2024

From Protein Structures to Clean Energy Materials to Cancer Therapies: Using AI to Understand and Exploit X-ray Damage Effects