Search for: All records

Implicit neural representations (INRs) have demonstrated success in a variety of applications, including inverse problems and neural rendering. An INR is typically trained to capture one signal of interest, resulting in learned neural features that are highly attuned to that signal. Assumed to be less generalizable, we explore the aspect of transferability of such learned neural features for fitting similar signals. We introduce a new INR training framework, STRAINER that learns transferrable features for fitting INRs to new signals from a given distribution, faster and with better reconstruction quality. Owing to the sequential layer-wise affine operations in an INR, we propose to learn transferable representations by sharing initial encoder layers across multiple INRs with independent decoder layers. At test time, the learned encoder representations are transferred as initialization for an otherwise randomly initialized INR. We find STRAINER to yield extremely powerful initialization for fitting images from the same domain and allow for ≈+10dB gain in signal quality early on compared to an untrained INR itself. STRAINER also provides a simple way to encode data-driven priors in INRs. We evaluate STRAINER on multiple in-domain and out-of-domain signal fitting tasks and inverse problems and further provide detailed analysis and discussion on the transferability of STRAINER's features. 

Simultaneous measurements of two nitrogen vacancy centers in diamond enables spatiotemporal magnetometry. 

Flying social insects can provide model systems for in-flight interactions in computationally-constrained aerial robot swarms. The social interactions in flying insects may be chemically modulated and quantified via recent measurement advancements able to simultaneously make precise measurements of insect wing and body motions. This paper presents the first in-flight quantitative measurements of ethanol-exposed honey bee body and wing kinematics in archival literature. Four high-speed cameras (9000 frames/sec) were used to record the wing and body motions of flying insects (Apis mellifera) and automated analysis was used to extract 9000 frame/sec measurements of honey bees’ wing and body motions through data association, hull reconstruction, and segmentation. The kinematic changes induced by exposure to incremental ethanol concentrations from 0% to 5% were studied using statistical analysis tools. Analysis considered trial-wise mean and maximum values and gross wingstroke parameters, and tested deviations for statistical significance using Welch’s t-test and Cohen’s d test. The results indicate a decrease in maximal heading and pitch rates of the body, and that roll rate is affected at high concentrations (5%). The wingstroke effects include a stroke frequency decrease and stroke amplitude increase for 2.5% or greater concentrations, gradual stroke inclination angle increase up to 2.5% concentration, and a more planar wingstroke with increasing concentration according to bulk wingstroke analysis. These ethanol-exposure effects provide a basis to separate ethanol exposure and neighbor effects in chemically mediated interaction studies. 

An intelligent system can provide sufficient collaborative opportunities and support yet fail to be pedagogically effective if the students are unwilling to participate. One of the common ways to assess motivation is using self-report questionnaires, which often do not take the context and the dynamic aspect of motivation into account. To address this, we propose personas, a user-centered design approach. We describe two design iterations where we: identify motivational factors related to students’ collaborative behaviors; and develop a set of representative personas. These personas could be embedded in an interface and be used as an alternative method to assess motivation within ITS. 

Computer-Supported Collaborative Learning (CSCL) environments are often designed to support collaboration within a single digital platform. However, with the growth of technology in classrooms, students often find themselves working in multiple contexts (i.e., a student might work face-to-face with a peer on one task and then move to engaging in an online discussion for homework). We have created a CSCL environment that aims to support student help-giving across a variety of digital platforms. This paper describes three cycles of a design-based research study that aims to design a system to support help-giving and improve interaction quantity and quality across different contexts as well as to better understand whether students benefit by the addition of multiple contexts. The paper shares major refinements across the three cycles that worked to balance research, pedagogical, and technological goals to improve students’ help-giving behavior in a middle-school mathematics classroom. 

A low-energy hardware implementation of deep belief network (DBN) architecture is developed using near-zero energy barrier probabilistic spin logic devices (p-bits), which are modeled to real- ize an intrinsic sigmoidal activation function. A CMOS/spin based weighted array structure is designed to implement a restricted Boltzmann machine (RBM). Device-level simulations based on precise physics relations are used to validate the sigmoidal relation between the output probability of a p-bit and its input currents. Characteristics of the resistive networks and p-bits are modeled in SPICE to perform a circuit-level simulation investigating the performance, area, and power consumption tradeoffs of the weighted array. In the application-level simulation, a DBN is implemented in MATLAB for digit recognition using the extracted device and circuit behavioral models. The MNIST data set is used to assess the accuracy of the DBN using 5,000 training images for five distinct network topologies. The results indicate that a baseline error rate of 36.8% for a 784x10 DBN trained by 100 samples can be reduced to only 3.7% using a 784x800x800x10 DBN trained by 5,000 input samples. Finally, Power dissipation and accuracy tradeoffs for probabilistic computing mechanisms using resistive devices are identified. 

Abstract The Pixel Luminosity Telescope is a silicon pixel detector dedicated to luminosity measurement at the CMS experiment at the LHC. It is located approximately 1.75 m from the interaction point and arranged into 16 “telescopes”, with eight telescopes installed around the beam pipe at either end of the detector and each telescope composed of three individual silicon sensor planes. The per-bunch instantaneous luminosity is measured by counting events where all three planes in the telescope register a hit, using a special readout at the full LHC bunch-crossing rate of 40 MHz. The full pixel information is read out at a lower rate and can be used to determine calibrations, corrections, and systematic uncertainties for the online and offline measurements. This paper details the commissioning, operational history, and performance of the detector during Run 2 (2015–18) of the LHC, as well as preparations for Run 3, which will begin in 2022.