Search for: All records

Learning with noisy labels (LNL) aims to train a high-performing model using a noisy dataset. We observe that noise for a given class often comes from a limited set of categories, yet many LNL methods overlook this. For example, an image mislabeled as a cheetah is more likely a leopard than a hippopotamus due to its visual similarity. Thus, we explore Learning with Noisy Labels with noise source Knowledge integration (LNL+K), which leverages knowledge about likely source(s) of label noise that is often provided in a dataset's meta-data. Integrating noise source knowledge boosts performance even in settings where LNL methods typically fail. For example, LNL+K methods are effective on datasets where noise represents the majority of samples, which breaks a critical premise of most methods developed for LNL. Our LNL+K methods can boost performance even when noise sources are estimated rather than extracted from meta-data. We provide several baseline LNL+K methods that integrate noise source knowledge into state-of-the-art LNL models that are evaluated across six diverse datasets and two types of noise, where we report gains of up to 23% compared to the unadapted methods. Critically, we show that LNL methods fail to generalize on some real-world datasets, even when adapted to integrate noise source knowledge, highlighting the importance of directly exploring LNL+K. 

Abstract Cardiac microtissues provide a promising platform for disease modeling and developmental studies, which require the close monitoring of the multimodal excitation-contraction dynamics. However, no existing assessing tool can track these multimodal dynamics across the live tissue. We develop a tissue-like mesh bioelectronic system to track these multimodal dynamics. The mesh system has tissue-level softness and cell-level dimensions to enable stable embedment in the tissue. It is integrated with an array of graphene sensors, which uniquely converges both bioelectrical and biomechanical sensing functionalities in one device. The system achieves stable tracking of the excitation-contraction dynamics across the tissue and throughout the developmental process, offering comprehensive assessments for tissue maturation, drug effects, and disease modeling. It holds the promise to provide more accurate quantification of the functional, developmental, and pathophysiological states in cardiac tissues, creating an instrumental tool for improving tissue engineering and studies. 

Most neural networks assume that input images have a fixed number of channels (three for RGB images). However, there are many settings where the number of channels may vary, such as microscopy images where the number of channels changes depending on instruments and experimental goals. Yet, there has not been a systemic attempt to create and evaluate neural networks that are invariant to the number and type of channels. As a result, trained models remain specific to individual studies and are hardly reusable for other microscopy settings. In this paper, we present a benchmark for investigating channel-adaptive models in microscopy imaging, which consists of 1) a dataset of varied-channel single-cell images, and 2) a biologically relevant evaluation framework. In addition, we adapted several existing techniques to create channel-adaptive models and compared their performance on this benchmark to fixed-channel, baseline models. We find that channel-adaptive models can generalize better to out-of-domain tasks and can be computationally efficient. We contribute a curated dataset and an evaluation API to facilitate objective comparisons in future research and applications. 

The Pacific Ocean region presents a significant gap in the equatorial coverage of the global Neutron Monitor (NM) network, hindering the detection of Solar Neutron Particles (SNP) and Galactic Cosmic Rays (GCR). To address this issue, we are redeploying the Haleakala Neutron Monitor (HLEA) on the island of Maui. HLEA was established in 1991 but was subsequently decommissioned in 2006 due to funding constraints. Its strategic location at a high altitude on Haleakala mountain, situated in the middle of the Pacific Ocean, offers unique advantages for SNP detection. The reinstatement of HLEA represents an invaluable opportunity to extend ground coverage for SNP and GCR detection, enhance the global NM network, and contribute to a deeper understanding of high-energy particle interactions. By harnessing the potential of this revitalized NM station, we aim to enrich space weather research and improve the efficacy of space weather monitoring systems, thereby enhancing our preparedness and resilience against space weather hazards.