Search for: All records

Energy-efficient image acquisition on the edge is crucial for enabling remote sensing applications where the sensor node has weak compute capabilities and must transmit data to a remote server/cloud for processing. To reduce the edge energy consumption, this paper proposes a sensor-algorithm co-designed system called SNAPPIX, which compresses raw pixels in the analog domain inside the sensor. We use coded exposure (CE) as the in-sensor compression strategy as it offers the flexibility to sample, i.e., selectively expose pixels, both spatially and temporally. SNAPPIX has three contributions. First, we propose a task-agnostic strategy to learn the sampling/exposure pattern based on the classic theory of efficient coding. Second, we co- design the downstream vision model with the exposure pattern to address the pixel-level non-uniformity unique to CE-compressed images. Finally, we propose lightweight augmentations to the image sensor hardware to support our in-sensor CE compres- sion. Evaluating on action recognition and video reconstruction, SNAPPIX outperforms state-of-the-art video-based methods at the same speed while reducing the energy by up to 15.4×. We have open-sourced the code at: https://github.com/horizon- research/SnapPix. 

Vision-language models are integral to computer vision research, yet many high-performing models remain closed-source, obscuring their data, design and training recipe. The research community has responded by using distillation from black-box models to label training data, achieving strong benchmark results, at the cost of measurable scientific progress. However, without knowing the details of the teacher model and its data sources, scientific progress remains difficult to measure. In this paper, we study building a Perception Language Model (PLM) in a fully open and reproducible framework for transparent research in image and video understanding. We analyze standard training pipelines without distillation from proprietary models and explore large-scale synthetic data to identify critical data gaps, particularly in detailed video understanding. To bridge these gaps, we release 2.8M human-labeled instances of fine-grained video question-answer pairs and spatio-temporally grounded video captions. Additionally, we introduce PLM-VideoBench, a suite for evaluating challenging video understanding tasks focusing on the ability to reason about "what", "where", "when", and "how" of a video. We make our work fully reproducible by providing data, training recipes, code & models. 

Permeability of binary mixtures of soils is important for several industrial and engineering applications. Previous models for predicting the permeability of a binary mixture of soils were primarily developed from Kozeny–Carman equation with an empirical approach. The permeability is predicted based on an equivalent particle size of the two species. This study is aimed to develop a model using a more fundamental approach. Instead of an equivalent particle size, the permeability is predicted based on the bimodal void sizes of the binary mixture. Because the bimodal void sizes are not available as commonly measured physical properties. We first develop an analytical method that has the capability of predicting the bimodal void sizes of a binary mixture. A permeability model is then developed based on the bimodal void sizes of the binary mixture. The developed permeability model is evaluated by comparing the predicted and experimentally measured results for binary mixtures of glass beads, crush sand, and gravel sand. The findings can contribute to a better understanding of the important influence of pore structure on the prediction of permeability. 

Datasets involving multivariate event streams are prevalent in numerous applications. We present a novel framework for modeling temporal point processes called clock logic neural networks (CLNN) which learn weighted clock logic (wCL) formulas as interpretable temporal rules by which some events promote or inhibit other events. Specifically, CLNN models temporal relations between events using conditional intensity rates informed by a set of wCL formulas, which are more expressive than related prior work. Unlike conventional approaches of searching for generative rules through expensive combinatorial optimization, we design smooth activation functions for components of wCL formulas that enable a continuous relaxation of the discrete search space and efficient learning of wCL formulas using gradient-based methods. Experiments on synthetic datasets manifest our model's ability to recover the ground-truth rules and improve computational efficiency. In addition, experiments on real-world datasets show that our models perform competitively when compared with state-of-the-art models. 

Abstract A systematic study on the biological effects of simulated microgravity (sµg) on human pluripotent stem cells (hPSC) is still lacking. Here, we used a fast-rotating 2-D clinostat to investigate the sµg effect on proliferation, self-renewal, and cell cycle regulation of hPSCs. We observed significant upregulation of protein translation of pluripotent transcription factors in hPSC cultured in sµg compared to cells cultured in 1g conditions. In addition to a significant increase in expression of telomere elongation genes. Differentiation experiments showed that hPSC cultured in sµg condition were less susceptible to differentiation compared to cells in 1g conditions. These results suggest that sµg enhances hPSC self-renewal. Our study revealed that sµg enhanced the cell proliferation of hPSCs by regulating the expression of cell cycle-associated kinases. RNA-seq analysis indicated that in sµg condition the expression of differentiation and development pathways are downregulated, while multiple components of the ubiquitin proteasome system are upregulated, contributing to an enhanced self-renewal of hPSCs. These effects of sµg were not replicated in human fibroblasts. Taken together, our results highlight pathways and mechanisms in hPSCs vulnerable to microgravity that imposes significant impacts on human health and performance, physiology, and cellular and molecular processes. 

ABSTRACT We present a revised analysis of the photometric reverberation mapping campaign of the narrow-line Seyfert 1 galaxy PKS 0558 − 504 carried out with the Swift Observatory during 2008–2010. Previously, Gliozzi et al. (2013) found using the Discrete Correlation Function (DCF) method that the short-wavelength continuum variations lagged behind variations at longer wavelengths, the opposite of the trend expected for thermal reprocessing of X-rays by the accretion disc, and they interpreted their results as evidence against the reprocessing model. We carried out new DCF measurements that demonstrate that the inverted lag-wavelength relationship found by Gliozzi et al. resulted from their having interchanged the order of the driving and responding light curves when measuring the lags. To determine the inter-band lags and uncertainties more accurately, we carried out new measurements with four independent methods. These give consistent results showing time delays increasing as a function of wavelength, as expected for the disc reprocessing scenario. The slope of the re-analysed delay spectrum appears to be roughly compatible with the predicted τ ∝ λ4/3 relationship for reprocessing by an optically thick and geometrically thin accretion disc, although the data points exhibit a large scatter about the fitted power-law trend.