While Convolutional Neural Networks (CNNs) have been widely successful in 2D human pose estimation, Vision Transformers (ViTs) have emerged as a promising alternative to CNNs, boosting state-of-the-art performance. However, the quadratic computational complexity of ViTs has limited their applicability for processing high-resolution images. In this paper, we propose three methods for reducing ViT’s computational complexity, which are based on selecting and processing a small number of most informative patches while disregarding others. The first two methods leverage a lightweight pose estimation network to guide the patch selection process, while the third method utilizes a set of learnable joint tokens to ensure that the selected patches contain the most important information about body joints. Experiments across six benchmarks show that our proposed methods achieve a significant reduction in computational complexity, ranging from 30% to 44%, with only a minimal drop in accuracy between 0% and 3.5%.
more »
« less
This content will become publicly available on January 18, 2025
Pattern recognition in the nucleation kinetics of non-equilibrium self-assembly
Inspired by biology’s most sophisticated computer, the brain, neural networks constitute a profound reformulation of computational principles1–3. Analogous high-dimensional, highly interconnected computational architectures also arise within information-processing molecular systems inside living cells, such as signal transduction cascades and genetic regulatory networks4–7. Might collective modes analogous to neural computation be found more broadly in other physical and chemical processes, even those that ostensibly play non-information-processing roles? Here we examine nucleation during self-assembly of multicomponent structures, showing that high-dimensional patterns of concentrations can be discriminated and classified in a manner similar to neural network computation. Specifically, we design a set of 917 DNA tiles that can self-assemble in three alternative ways such that competitive nucleation depends sensitively on the extent of colocalization of high-concentration tiles within the three structures. The system was trained in silico to classify a set of 18 grayscale 30 × 30 pixel images into three categories. Experimentally, fluorescence and atomic force microscopy measurements during and after a 150 hour anneal established that all trained images were correctly classified, whereas a test set of image variations probed the robustness of the results. Although slow compared to previous biochemical neural networks, our approach is compact, robust and scalable. Our findings suggest that ubiquitous physical phenomena, such as nucleation, may hold powerful information-processing capabilities when they occur within high-dimensional multicomponent systems.
more » « less- Award ID(s):
- 2011854
- PAR ID:
- 10507276
- Publisher / Repository:
- Springer Nature
- Date Published:
- Journal Name:
- Nature
- Volume:
- 625
- Issue:
- 7995
- ISSN:
- 0028-0836
- Page Range / eLocation ID:
- 500 to 507
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
more » « less
Processing social information from faces is difficult for individuals with autism spectrum disorder (ASD). However, it remains unclear whether individuals with ASD make high-level social trait judgments from faces in the same way as neurotypical individuals. Here, we comprehensively addressed this question using naturalistic face images and representatively sampled traits. Despite similar underlying dimensional structures across traits, online adult participants with self-reported ASD showed different judgments and reduced specificity within each trait compared with neurotypical individuals. Deep neural networks revealed that these group differences were driven by specific types of faces and differential utilization of features within a face. Our results were replicated in well-characterized in-lab participants and partially generalized to more controlled face images (a preregistered study). By investigating social trait judgments in a broader population, including individuals with neurodevelopmental variations, we found important theoretical implications for the fundamental dimensions, variations, and potential behavioral consequences of social cognition.
-
Assembly of complex structures from a small set of tiles is a common theme in biology. For example, many copies of identical proteins make up polyhedron-shaped, viral capsids and tubulin can make long microtubules. This inspired the development of tile-based DNA self-assembly for nanoconstruction, particularly for structures with high symmetries. In the final structure, each type of motif will adopt the same conformation, either rigid or with defined flexibility. For structures that have no symmetry, their assembly remains a challenge from a small set of tiles. To meet this challenge, algorithmic self-assembly has been explored driven by computational science, but it is not clear how to implement this approach to one-dimensional (1D) structures. Here, we have demonstrated that a constant shift of a conformational equilibrium could allow 1D structures to evolve. As shown by atomic force microscopy imaging, one type of DNA tile successfully assembled into DNA spirals and concentric circles, which became less and less curved from the structure's center outward. This work points to a new direction for tile-based DNA assembly.more » « less
-
Interactive visual analytics over distributed systems housing voluminous datasets is hindered by three main factors - disk and network I/O, and data processing overhead. Requests over geospatial data are prone to erratic query load and hotspots due to users’ simultaneous interest over a small sub-domain of the overall data space at a time. Interactive analytics in a distributed setting is further hindered in cases of voluminous datasets with large/high-dimensional data objects, such as multi-spectral satellite imagery. The size of the data objects prohibits efficient caching mechanisms that could significantly reduce response latencies. Additionally, extracting information from these large data objects incurs significant data processing overheads and they often entail resource-intensive computational methods. Here, we present our framework, ARGUS, that extracts low- dimensional representation (embeddings) of high-dimensional satellite images during ingestion and houses them in the cache for use in model-driven analysis relating to wildfire detection. These embeddings are versatile and are used to perform model- based extraction of analytical information for a set of dif- ferent scenarios, to reduce the high computational costs that are involved with typical transformations over high-dimensional datasets. The models for each such analytical process are trained in a distributed manner in a connected, multi-task learning fashion, along with the encoder network that generates the original embeddings.more » « less
-
As one of the popular deep learning methods, deep convolutional neural networks (DCNNs) have been widely adopted in segmentation tasks and have received positive feedback. However, in segmentation tasks, DCNN-based frameworks are known for their incompetence in dealing with global relations within imaging features. Although several techniques have been proposed to enhance the global reasoning of DCNN, these models are either not able to gain satisfying performances compared with traditional fully-convolutional structures or not capable of utilizing the basic advantages of CNN-based networks (namely the ability of local reasoning). In this study, compared with current attempts to combine FCNs and global reasoning methods, we fully extracted the ability of self-attention by designing a novel attention mechanism for 3D computation and proposed a new segmentation framework (named 3DTU) for three-dimensional medical image segmentation tasks. This new framework processes images in an end-to-end manner and executes 3D computation on both the encoder side (which contains a 3D transformer) and the decoder side (which is based on a 3D DCNN). We tested our framework on two independent datasets that consist of 3D MRI and CT images. Experimental results clearly demonstrate that our method outperforms several state-of-the-art segmentation methods in various metrics.more » « less
