Search for: All records

Machine unlearning is becoming increasingly important as deep models become more prevalent, particularly when there are frequent requests to remove the influence of specific training data due to privacy concerns or erroneous sensing signals. Spatial-temporal Graph Neural Networks, in particular, have been widely adopted in real-world applications that demand efficient unlearning, yet research in this area remains in its early stages. In this paper, we introduce STEPS, a framework specifically designed to address the challenges of spatio-temporal graph unlearning. Our results demonstrate that STEPS not only ensures data continuity and integrity but also significantly reduces the time required for unlearning, while minimizing the accuracy loss in the new model compared to a model with 0% unlearning. 

Functional thin coatings are crucial in modern and emerging technologies, providing specified surface properties and protection, thereby influencing the performance and lifetime of materials and devices. The electrodeposition of polymer networks (EPoN) has recently been reported as a facile and potentially broadly applicable method to fabricate conformal polymeric ultrathin films on conductive substrates with arbitrary shapes and surface topography under mild solution conditions. In this work, a new generation of EPoN is introduced that utilizes a chemically reactive polymer appended by a small fraction of a electrochemical crosslinker as side groups. This EPoN iteration eliminates the need for precise end-group functionalization, enables the tuning of crosslink density and film thickness independent of polymer size, and the resulting reactive ultrathin films are amenable to post-deposition modification with desired functionalities using facile click-chemistry. To demonstrate this concept, we electrodeposit polyisoprene with small side-group fractions of the oxidative crosslinker phenol (<5%) as a thiol–ene-reactive polymer-network coating. The EPoN-derived ultrathin films are tunable and uniform with a thickness in the 100s of nanometers depending on phenol fraction and electrodeposition potential, and show a conformal morphology on complex porous electrode architectures. We further demonstrate post-EPoN functionalization of the ultrathin polyisoprene coatings with thiol–ene chemistry. 

Advances in precision coatings are critical in enhancing the functionality of porous materials and the performance of three-dimensionally (3-D) micro-architected devices in applications ranging from molecular sorption and separation to energy storage and conversion. To address this need, we report the cathodic electrodeposition of polymer networks (EPoN) that utilizes the coupling between pre-synthesized polymers with electrochemically active end groups and a complementary crosslinker to form a step-growth polymer network. The electrochemically mediated crosslinking reaction confines the network formation to the electrode surface in a passivating and self-limiting film growth, preventing uncontrolled precipitation and deposition away from the surface. The cathodic electrodeposition is compatible with a variety of conductive substrates, which is demonstrated for 3-D carbons and metals with micron-scale pores of high aspect ratio. The entire pore surface of the 3-D electrodes is enveloped by a conformal polymer thin film that is free of detectable defects and highly electronically insulating for its potential use as an ultrathin artificial electrolyte interphase or solid polymer electrolyte. Since our EPoN concept decouples the polymer functionality from its electrodeposition chemistry, we envision it to be a widely applicable method to coat various conductive non-planar and micro-architected 3-D substrates with polymers of broad functionalities. 

Spatio-temporal deep learning has drawn a lot of attention since many downstream real-world applications can benefit from accurate predictions. For example, accurate prediction of heavy rainfall events is essential for effective urban water usage, flooding warning, and mitigation. In this paper, we propose a strategy to leverage spatially connected real-world features to enhance prediction accuracy. Specifically, we leverage spatially connected real-world climate data to predict heavy rainfall risks in a broad range in our case study. We experimentally ascertain that our Trans-Graph Convolutional Network (TGCN) accurately predicts heavy rainfall risks and real estate trends, demonstrating the advantage of incorporating external spatially-connected real-world data to improve model performance, and it shows that this proposed study has a significant potential to enhance spatio-temporal prediction accuracy, aiding in efficient urban water usage, flooding risk warning, and fair housing in real estate. 

Multifunctional thin films in energy-related devices often must be electrically insulating where a single nanoscale defect can result in complete device-scale failure. Locating and characterizing such defects presents a fundamental problem where high-resolution imaging methods are needed to find defects, but imaging with high spatial resolution limits the field of view and thus the measurement throughput. Here, we present a novel high-throughput method for detecting sub-micron defects in insulating thin films by leveraging the electrochemiluminescence (ECL) of luminol. Through a systematic study of reagent concentrations, buffers, voltage, and excitation time, we identify optimized conditions at which it is possible to detect sub-micron defects at high-throughput. Extrapolating from the signal to background observed for detecting 440 nm wide lines and 620 nm diameter circles, we estimate the minimum detectable features to be lines as narrow as 2.5 nm in width and pinholes as small as 70 nm in radius. We further explore this method by using it to characterize a nominally insulating phenol film and find conductive defects that are cross-correlated with high-resolution atomic force microscopy to provide feedback to synthesis. Given this assay’s inherent parallelizability and scalability, it is expected to have a major impact on the automated discovery of multifunctional films.