Search NASA⌕ Search

SEARCH · Search NASA

Results for “ATTENTION”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 127 records · Page 7
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Extending Symmetry-Preserving Attention Networks (SPANet) for Jet Assignment in Fully Hadronic \(t\bar{t}\) Events

Fully hadronic \(t\bar{t}\) reconstruction requires assigning reconstructed jets to the two top-quark decay branches, \(t\bar{t}\to(Wb)(Wb)\to(qqb)(qqb)\). This is a combinatorially large, symmetry-rich, and frequently underconstrained set-assignment problem due to detector effects and partial reconstructability. SPANet provides a strong symmetry-preserving baseline for this task, but its standard inference eagerly commits to a single hypothesis and does not explicitly treat branch-level reconstructability decisions. We present a two-stage extension of the SPANet pipeline for fully hadronic \(t\bar{t}\) events: (i) SPANet is modified into a proposal model that generates a shortlist of candidate jet assignments; (ii) a custom dual-head set transformer is trained to re-rank these candidates and identify reconstructible branches. Our extension gives modest improvements in candidate-selection efficiency, with SPANet achieving \(69.9\%\) and our extension achieving \(71.1\%\). Finally, an oracle study of the generated shortlist shows a substantial upper bound of \(88.5\%\) on the evaluated fully reconstructable subset, suggesting that the shortlist contains significant residual information that is not fully exploited.

Lisondodi Tada, Mateo [Puerto Rico U., Mayaguez] (↗
DOE OSTI2023

Tyramine and its Amtyr1 receptor modulate attention in honey bees (Apis mellifera)

Animals must learn to ignore stimuli that are irrelevant to survival and attend to ones that enhance survival. When a stimulus regularly fails to be associated with an important consequence, subsequent excitatory learning about that stimulus can be delayed, which is a form of nonassociative conditioning called ‘latent inhibition’. Honey bees show latent inhibition toward an odor they have experienced without association with food reinforcement. Moreover, individual honey bees from the same colony differ in the degree to which they show latent inhibition, and these individual differences have a genetic basis. To investigate the mechanisms that underly individual differences in latent inhibition, we selected two honey bee lines for high and low latent inhibition, respectively. We crossed those lines and mapped a Quantitative Trait Locus for latent inhibition to a region of the genome that contains the tyramine receptor gene Amtyr1 [We use Amtyr1 to denote the gene and AmTYR1 the receptor throughout the text.]. We then show that disruption of Amtyr1 signaling either pharmacologically or through RNAi qualitatively changes the expression of latent inhibition but has little or slight effects on appetitive conditioning, and these results suggest that AmTYR1 modulates inhibitory processing in the CNS. Electrophysiological recordings from the brain during pharmacological blockade are consistent with a model that AmTYR1 indirectly regulates at inhibitory synapses in the CNS. Our results therefore identify a distinct Amtyr1 -based modulatory pathway for this type of nonassociative learning, and we propose a model for how Amtyr1 acts as a gain control to modulate hebbian plasticity at defined synapses in the CNS. We have shown elsewhere how this modulation also underlies potentially adaptive intracolonial learning differences among individuals that benefit colony survival. Finally, our neural model suggests a mechanism for the broad pleiotropy this gene has on several different behaviors.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Interpreting and Accelerating Transformers for Jet Tagging

Attention-based transformers are ubiquitous in machine learning applications from natural language processing to computer vision. In high energy physics, one central application is to classify collimated particle showers in colliders based on the particle of origin, known as jet tagging. In this work, we study the interpretatbility and prospects for acceleration of Particle Transformer (ParT), a state-of-the-art model, leverages particle-level attention to improve jet-tagging performance. We analyzing ParT's attention maps and particle-pair correlations in the eta-phi plane, revealing intriguing features, such as a binary attention pattern that identifies critical substructure in jets. These insights enhance our understanding of the model's internal workings and learning process and hint at ways to improve its efficiency. Along these lines, we also explore low-rank attention, attention alternatives, and dynamic quantization to accelerate transformers for jet tagging. With quantization, we achieve a 50% reduction in model size and a 10% increase in inference speed without compromising accuracy. These combined efforts enhance both the performance and the interpretability of transformers in high-energy physics, opening avenues for more efficient and physics-driven model designs.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

Identification of Blue Horizontal Branch Stars with Multimodal Fusion

Blue Horizontal Branch stars (BHBs) are ideal tracers to probe the global structure of the milky Way (MW), and the increased size of the BHB star sample could be helpful to accurately calculate the MW's enclosed mass and kinematics. Large survey telescopes have produced an increasing number of astronomical images and spectra. However, traditional methods of identifying BHBs are limited in dealing with the large scale of astronomical data. A fast and efficient way of identifying BHBs can provide a more significant sample for further analysis and research. Therefore, in order to fully use the various data observed and further improve the identification accuracy of BHBs, we have innovatively proposed and implemented a Bi-level attention mechanism-based Transformer multimodal fusion model, called Bi-level Attention in the Transformer with Multimodality (BATMM). The model consists of a spectrum encoder, an image encoder, and a Transformer multimodal fusion module. The Transformer enables the effective fusion of data from two modalities, namely image and spectrum, by using the proposed Bi-level attention mechanism, including cross-attention and self-attention. As a result, the information from the different modalities complements each other, thus improving the accuracy of the identification of BHBs. The experimental results show that the F1 score of the proposed BATMM is 94.78%, which is 21.77% and 2.76% higher than the image and spectral unimodality, respectively. It is therefore demonstrated that higher identification accuracy of BHBs can be achieved by means of using data from multiple modalities and employing an efficient data fusion strategy.

79 ASTRONOMY AND ASTROPHYSICS↗