Kaikki aineistot
Lisää
Abstract Benefited from the powerful ability of spatial temporal Graph Convolutional Networks (ST-GCNs), skeleton-based human action recognition has gained promising success. However, the node interaction through message propagation does not always provide complementary information. Instead, it May even produce destructive noise and thus make learned representations indistinguishable. Inevitably, the graph representation would also become over-smoothing especially when multiple GCN layers are stacked. This paper proposes spatial-temporal graph deconvolutional networks (ST-GDNs), a novel and flexible graph deconvolution technique, to alleviate this issue. At its core, this method provides a better message aggregation by removing the embedding redundancy of the input graphs from either node-wise, frame-wise or element-wise at different network layers. Extensive experiments on three current most challenging benchmarks verify that ST-GDN consistently improves the performance and largely reduce the model size on these datasets.
Abstract Graph Convolutional Networks (GCNs) have already demonstrated their powerful ability to model the irregular data, e.g., skeletal data in human action recognition, providing an exciting new way to fuse rich structural information for nodes residing in different parts of a graph. In human action recognition, current works introduce a dynamic graph generation mechanism to better capture the underlying semantic skeleton connections and thus improves the performance. In this paper, we provide an orthogonal way to explore the underlying connections. Instead of introducing an expensive dynamic graph generation paradigm, we build a more efficient GCN on a Riemann manifold, which we think is a more suitable space to model the graph data, to make the extracted representations fit the embedding matrix. Specifically, we present a novel spatial-temporal GCN (ST-GCN) architecture which is defined via the Poincaré geometry such that it is able to better model the latent anatomy of the structure data. To further explore the optimal projection dimension in the Riemann space, we mix different dimensions on the manifold and provide an efficient way to explore the dimension for each ST-GCN layer. With the final resulted architecture, we evaluate our method on two current largest scale 3D datasets, i.e., NTU RGB+D and NTU RGB+D 120. The comparison results show that the model could achieve a superior performance under any given evaluation metrics with only 40% model size when compared with the previous best GCN method, which proves the effectiveness of our model.
Abstract Skeleton based action recognition is playing a critical role in computer vision research, its applications have been widely deployed in many areas. Currently, benefiting from the graph convolutional networks (GCN), the performance of this task is dramatically improved due to the powerful ability of GCN for modeling the Non-Euclidean data. However, most of these works are designed for the clean skeleton data while one unavoidable drawback is such data is usually noisy in reality, since most of such data is obtained using depth camera or even estimated from RGB camera, rather than recorded by the high quality but extremely costly Motion Capture (MoCap) [1] system. Under this circumstance, we propose a novel GCN framework with adversarial training to deal with the noisy skeleton data. With the guiding of the clean data in the semantic level, a reliable graph embedding can be extracted for noisy skeleton data. Besides, a discriminator is introduced such that the feature representation could further improved since it is learned with an adversarial learning fashion. We empirically demonstrate the proposed framework based on two current largest scale skeleton-based action recognition datasets. Comparison results show the superiority of our method when compared to the state-of-the-art methods under the noisy settings.
Abstract The skeletal data has been an alternative for the human action recognition task as it provides more compact and distinct information compared to the traditional RGB input. However, unlike the RGB input, the skeleton data lies in a non-Euclidean space that traditional deep learning methods are not able to use their fullest potential. Fortunately, with the emerging trend of Geometric deep learning, the spatial-temporal graph convolutional network (ST-GCN) has been proposed to deal with the action recognition problem from skeleton data. ST-GCN and its variants fit well with skeleton-based action recognition and are becoming the mainstream frameworks for this task. However, the efficiency and the performance of the task are hindered by either fixing the skeleton joint correlations or providing a computational expensive strategy to construct a dynamic topology for the skeleton. We argue that many of these operations are either unnecessary or even harmful for the task. By theoretically and experimentally analysing the state-of-the-art ST-GCNs, we provide a simple but efficient strategy to capture the global graph correlations and thus efficiently model the representation of the input graph sequences. Moreover, the global graph strategy also reduces the graph sequence into the Euclidean space, thus a multi-scale temporal filter is introduced to efficiently capture the dynamic information. With the method, we are not only able to better extract the graph correlations with much fewer parameters (only 12.6% of the current best), but we also achieve a superior performance. Extensive experiments on current largest 3D datasets, NTU-RGB+D and NTU-RGB+D 120, demonstrate the ability of our network to perform efficient and lightweight priority on this task.
Abstract Recently, hyperbolic deep neural networks (HDNNs) have been gaining momentum as the deep representations in the hyperbolic space provide high fidelity embeddings with few dimensions, especially for data possessing hierarchical structure. Such a hyperbolic neural architecture is quickly extended to many different scientific fields, including natural language processing, single-cell RNA-sequence analysis, graph embedding, financial analysis, and computer vision. The promising results demonstrate its superior capability, significant compactness of the model, and a substantially better physical interpretability than its counterpart in the Euclidean space. To stimulate future research, this paper presents a coherent and a comprehensive review of the literature around the neural components in the construction of HDNN, as well as the generalization of the leading deep approaches to the hyperbolic space. It also presents current applications of various tasks, together with insightful observations and identifying open questions and promising future directions.
Abstract In this work, the Ti-48Al-2Nb-2Cr (at. %) alloy was successfully brazed using a Cu-based amorphous filler in 600 s under varied brazing temperatures. The element diffusion, microstructure, and precipitation phase of the joints are analyzed in detail, and the formation schemes are discussed. Reaction products in the joints are found as AlCuTi, Ti2Al, α-Ti, and (Ti,Zr)2(Cu,Ni). The interfacial microstructures varied subjected to the brazing temperature, while the shear strength of the joint firstly increased, and then accordingly decreased. The maximum shear strength of 266 MPa was reached under a brazing temperature of 1213 K and a holding time of 600 s. A formation mechanism was proposed to explain the shear strength variation following the width and amount of brittle compounds in the interfacial reaction layer.
Abstract Emotion recognition from body gestures is challenging since similar emotions can be expressed by arbitrary spatial configurations of joints, which results in relying on modeling spatial-temporal patterns from a more global level. However, most recent powerful graph convolution networks (GCNs) separate the spatial and temporal modeling into isolated processes, where GCN models spatial interactions using partially fixed adjacent matrices and 1D convolution captures temporal dynamics, which is insufficient for emotion recognition. In this work, we propose the 3D-Shift GCN, which enables interactions of joints within a spatial-temporal volume for global feature extraction. Besides, we further develop a multiscale architecture, the MS-Shift GCN, to fuse features captured under different temporal ranges for modeling richer dynamics. After conducting evaluation on two regular action recognition benchmarks and two gesture based emotion recognition datasets, the results show that the proposed method outperforms several state-of-the-art methods.
Abstract With the strength of deep generative models, 3D pose transfer regains intensive research interests in recent years. Existing methods mainly rely on a variety of constraints to achieve the pose transfer over 3D meshes, e.g., the need for manually encoding for shape and pose disentanglement. In this paper, we present an unsupervised approach to conduct the pose transfer between any arbitrate given 3D meshes. Specifically, a novel Intrinsic-Extrinsic Preserved Generative Adversarial Network (IEP-GAN) is presented for both intrinsic (i.e., shape) and extrinsic (i.e., pose) information preservation. Extrinsically, we propose a co-occurrence discriminator to capture the structural/pose invariance from distinct Laplacians of the mesh. Meanwhile, intrinsically, a local intrinsic-preserved loss is introduced to preserve the geodesic priors while avoiding heavy computations. At last, we show the possibility of using IEP-GAN to manipulate 3D human meshes in various ways, including pose transfer, identity swapping and pose interpolation with latent code vector arithmetic. The extensive experiments on various 3D datasets of humans, animals and hands qualitatively and quantitatively demonstrate the generality of our approach. Our proposed model produces better results and is substantially more efficient compared to recent state-of-the-art methods. Code is available: https://github.com/mikecheninoulu/Unsupervised_IEPGAN
Abstract The high temperature creep behaviors of a Zr-based bulk metallic glass (BMG) are studied by uniaxial tensile creep experiments under applied stresses of 50–180 MPa at temperatures of 660–700 K. The microstructural observations of the BMG samples after creep tests show that crystalline phases can be detected under high temperature or high applied stress. Constitutive models for predicting the high temperature creep behaviors of the studied Zr-based BMG are established based on the Ɵ projection method. The creep activation energy and stress exponent are also calculated to establish the creep model. The parameters of the established models are found to be closely associated with the applied stress and temperature. The results show an excellent agreement between the measured and predicted results, confirming the validity of the established model to accurately estimate the high temperature creep curves for the Zr-based BMG. Moreover, based on the classical diffusion creep theory, a schematic model is proposed to describe the creep behaviors of BMGs from the framework of free volume theory.
Abstract Gifted with unique optical and hydrophobic properties, the plant leaves have been recently considered as micro/nanostructure prototypes for functional surface engineering. Imprinting bio-inspired structures onto surfaces can yield in similar functional properties than in the nature. In this article, we report on a simple and effective method to copy leaf surface structures onto poly-(methyl methacrylate) sheets. The replicated surface structures reduce optical reflectance and enhance optical haze. Besides, the artificial polymer sheets exhibit good hydrophobic properties. Correlation between optical haze and hydrophobicity was studied.
Abstract The naturally evolved sunlight harvesters are not limited to foliage. Animals also harvest sunlight for light-heat conversion. A typical antireflective and light-trapping scheme has been well demonstrated on thin butterfly scales where solar energy is converted to heat besides being diffracted for surface coloration. Biomimicking scale structures offers a unique route to enhance light harvesting efficiencies happening on manmade solar cells. Herein, we performed a computational investigation of using microstructures on black butterflies for solar cell efficiency enhancement. Scale microstructures were obtained from nine species of black butterflies and employed as coating structures in numerical models built on Si-slabs. Introducing butterfly wing structures not only reduces the light reflection and transmittance but also increases the light absorption within Si-slabs. Surface reflection was decreased down to 10%, and the short-circuit current was increased by 66% correspondingly. An antireflection design strategy is given and hoped to benefit light harvesting in Si-based solar cells eventually.
Abstract As an improvement over organic or inorganic layered crystals, the synthetic monolayer ZnO(M) inherits semiconductivity and hostability from its bulk, yet it acts as a promising host for dilute magnetic semiconductors. Here, we report the electronic and magnetic properties of ZnO(M) doped with one 3d transition metal ion and simultaneously adsorbed with another 3d transition metal ion. Two sequences are studied, one where the dopant is fixed to Mn and the adsorbate is varied from Sc to Zn and another where the dopant and adsorbate are reversed. First-principles results show that the stable adsorbed−doped systems possess a lower bandgap energy than that of the host. System magnetic moments can be tuned to |5 − x|μB, where x refers to the magnetic moment of the individual 3d atom. An interplay between superexchange and direct exchange yields a ferromagnetic system dually adsorbed−doped with Mn. In addition to a novel material design route, the magnetic interaction mechanism is found beyond two dimensions, having been identified for its three-dimensional bulk and zero-dimensional cluster counterparts.
Abstract The microstructure, mechanical properties, and corrosion resistance to simulated body fluid solution behavior of as-extruded Mg-1.8Zn-0.5Zr alloys with different Gd additions are investigated. It is found that dynamic recrystallization occurs in the alloys during extrusion and the grain size gradually decreases with Gd alloying. The mechanical properties and corrosion resistance to simulated body fluid of the investigated alloys enhance firstly and then weaken with the increase in Gd content. The results reveal that the Mg-1.8Zn-0.5Zr with a 1.5 wt.% Gd addition has optimized mechanical properties and corrosion resistance. A three-stage corrosion mechanism, including sequential stages from hydroxidation, phosphatization and hydroxidation, to formation-dissolution dynamic equilibrium, is proposed through electrochemical measurements and corroded surface analyses. This study reveals the extruded Mg-1.8Zn-0.5Zr-1.5Gd alloy can be regarded as a potential candidate for using as biodegradable magnesium implants.
Abstract Direct sunlight-induced water splitting for photocatalytic hydrogen evolution is the dream for an ultimate clean energy source. So far, typical photocatalysts require complicated synthetic processes and barely work without additives or electrolytes. Here, we report the realization of a hydrogen evolution strategy with a novel Ni–Ag–MoS2 ternary nanocatalyst under visible/sun light. Synthesized through an ultrasound-assisted wet method, the composite exhibits stable catalytic activity for long-term hydrogen production from both pure and natural water. A high efficiency of 73 μmol g−1 W−1 h−1 is achieved with only a visible light source and the (MoS2)84Ag10Ni6 catalyst, matching the values of present additive-enriched photocatalysts. Verified by experimental characterizations and first-principles calculations, the enhanced photocatalytic ability is attributed to effective charge migration through the dangling bonds at the Ni–Ag–MoS2 alloy interface and the activation of the MoS2 basal planes.
Abstract Construction of heterojunctions is conventionally regarded as the prevailing technique to enhance solar-driven photocatalytic water splitting and photodegradation of pollutants. Herein, we report a novel design of a ternary Bi2O3/Bi/ZnIn2S4 system, which was facilely synthesized to satisfy these stringent criteria for sunlight photocatalytic removal of organic and ionic pollutants and hydrogen evolution. Bi2O3/Bi/ZnIn2S4 could degrade 2,4-dinitrophenol (94.6%), tetracycline (96.5%), and Cr6+ (96.3%) effectively under visible light and give a hydrogen production rate of 482.5 μmol·g−1·h−1 under visible light. Based on first-principles calculations and electrochemical results, our system could be identified as a Z-scheme. Photocorrosion of the sulfide is prohibited while the catalytic capabilities are simultaneously benefited due to lowered bandgap in light harvesting, internal electric fields in charge separations, and surface plasmonic resonance enhanced electron boost.
Abstract Accurate knowledge of the oxidation stages of lithium is crucially important for developing next-generation Li-air batteries. The intermediate oxidation stages, however, differ in the bulk and cluster forms of lithium. In this letter, using first-principles calculations, we predict several reaction pathways leading to the formation of Li3O+ superatoms. Experimental results based on time-of-flight mass spectrometry and laser ablation of oxidized lithium bulk samples agreed well with our theoretical calculations. Additionally, the highest occupied molecular orbital-lowest unoccupied molecular orbital gap of Li3O+ was close to the energy released in one of these reaction paths, indicating that the superatom could act as a candidate charge-discharge unit.
Abstract In this work, we present laser coloration on 304 stainless steel using nanosecond laser. Surface modifications are tuned by adjusting laser parameters of scanning speed, repetition rate, and pulse width. A comprehensive study of the physical mechanism leading to the appearance is presented. Microscopic patterns are measured and employed as input to simulate light-matter interferences, while chemical states and crystal structures of composites to figure out intrinsic colors. Quantitative analysis clarifies the final colors and RGB values are the combinations of structural colors and intrinsic colors from the oxidized pigments, with the latter dominating. Therefore, the engineering and scientific insights of nanosecond laser coloration highlight large-scale utilization of the present route for colorful and resistant steels.
Abstract Pioneering explorations of the two-dimensional (2D) inorganic layered crystals (ILCs) in electronics have boosted low-dimensional materials research beyond the prototypical but semi-metallic graphene. Thanks to species variety and compositional richness, ILCs are further activated as hosting matrices to reach intrinsic magnetism due to their semiconductive natures. Herein, we briefly review the latest progresses of manipulation strategies that introduce magnetism into the nonmagnetic 2D and quasi-2D ILCs from the first-principles computational perspectives. The matrices are concerned within naturally occurring species such as MoS2, MoSe2, WS2, BN, and synthetic monolayers such as ZnO and g-C2N. Greater attention is spent on nondestructive routes through magnetic dopant adsorption; defect engineering; and a combination of doping-absorbing methods. Along with structural stability and electric uniqueness from hosts, tailored magnetic properties are successfully introduced to low-dimensional ILCs. Different from the three-dimensional (3D) bulk or zero-dimensional (0D) cluster cases, origins of magnetism in the 2D space move past most conventional physical models. Besides magnetic interactions, geometric symmetry contributes a non-negligible impact on the magnetic properties of ILCs, and surprisingly leads to broken symmetry for magnetism. At the end of the review, we also propose possible combination routes to create 2D ILC magnetic semiconductors, tentative theoretical models based on topology for mechanical interpretations, and next-step first-principles research within the domain.
Abstract The topic of durable coloration and passivation of metal surfaces using state-of-the-art techniques has gained enormous attention and devotion with unremitting efforts of researchers worldwide. Although femtosecond laser marking has been performed on many metals, the related coloration mechanisms are mainly referred to structural colors produced by the interaction of visible light with periodic surface structures. Yet, general quantitative determination of the resulting colors and their origins remain elusive. In this work, we realized quantitative separations of structural colors and compositional pigmentary colors on 301LN austenitic stainless steel surfaces that were treated by femtosecond laser machining. The overall color information was extracted from surface reflectance, with structural color given by numerical simulations, and oxide compositions by chemical state analysis. It was shown that the laser-induced apparent colors of 301LN steel surfaces were combinations of structural and compositional colorations, with the former dominating the angular response and the latter setting up the brownish bases. In addition to the quantification of colors, the analysis method in this work may be useful for the generation and specification of tailored color palettes for practical coloration on metal surfaces by femtosecond laser marking.
Abstract Background: Haemophilus influenzae (H. influenzae), Streptococcus pneumoniae (pneumococcus) and influenza vaccines are administered in children to prevent infections caused by these pathogens. The benefits of vaccination for asthma control in children and the elicited immune response are not fully understood. This study aimed to investigate the impact of these vaccinations on respiratory infections, asthma symptoms, asthma severity and control status, pathogen colonization and in vitro immune responses to different stimulants mimicking infections in asthmatic children. Methods: Children aged 4–6 years were recruited into the multicentre prospective PreDicta study conducted across five European countries. Information about vaccination history, infections, antibiotic use, inhaled corticosteroid (ICS) use and asthma symptoms in the last 12 months were obtained from questionnaires of the study. Nasopharyngeal samples were collected at the first visit to assess bacterial and viral colonization, and venous blood for isolation of peripheral blood mononuclear cells (PBMCs). The PBMCs were stimulated with phytohemagglutinin, R848, Poly I:C and zymosan. The levels of 22 cytokines and chemokines were measured in cell culture supernatants using a luminometric multiplex assay. Results: One-hundred and forty asthmatic preschool children (5.3 ± 0.7 years) and 53 healthy children (5.0 ± 0.8 years) from the PreDicta cohort were included in the current study. Asthmatic children were associated with more frequent upper and lower respiratory infections, and more frequent and longer duration of antibiotic use compared with healthy children. In asthmatic children, sufficient H. influenzae vaccination was associated with a shorter duration of upper respiratory infection (URI) and overall use and average dose of ICS. The airway colonization was characterized by less pneumococcus and more rhinovirus. Pneumococcal vaccination was associated with a reduction in the use rate and average dose of ICS, improved asthma control, and less human enterovirus and more H. influenzae and rhinovirus (RV) airway colonization. Influenza vaccination in the last 12 months was associated with a longer duration of URI, but with a decrease in the occurrence of lower respiratory infection (LRI) and the duration of gastrointestinal (GI) infection and antibiotic use. Asthmatic preschoolers vaccinated with H. influenzae, pneumococcus or influenza presented higher levels of Th1-, Th2-, Th17- and regulatory T cells (Treg)-related cytokines in unstimulated PBMCs. Under stimulation, PBMCs from asthmatic preschoolers with pneumococcal vaccination displayed a predominant anti-inflammatory immune response, whereas PBMCs from asthmatic children with sufficient H. influenzae or influenza vaccination were associated with both pro- and anti-inflammatory immune responses. Conclusion: In asthmatic preschoolers, the standard childhood vaccinations to common respiratory pathogens have beneficial effects on asthma control and may modulate immune responses relevant to asthma pathogenesis.
Abstract A critical factor for electronics based on inorganic layered crystals stems from the electrical contact mode between the semiconducting crystals and the metal counterparts in the electric circuit. Here, a materials tailoring strategy via nanocomposite decoration is carried out to reach metallic contact between MoS2 matrix and transition metal nanoparticles. Nickel nanoparticles (NiNPs) are successfully joined to the sides of a layered MoS2 crystal through gold nanobuffers, forming semiconducting and magnetic NiNPs@MoS2 complexes. The intrinsic semiconducting property of MoS2 remains unchanged, and it can be lowered to only few layers. Chemical bonding of the Ni to the MoS2 host is verified by synchrotron radiation based photoemission electron microscopy, and further proved by first‐principles calculations. Following the system’s band alignment, new electron migration channels between metal and the semiconducting side contribute to the metallic contact mechanism, while semiconductor–metal heterojunctions enhance the photocatalytic ability.
Advances in the synthesis and scalable manufacturing of single-walled carbon nanotubes (SWCNTs) remain critical to realizing many important commercial applications. Here we review recent breakthroughs in the synthesis of SWCNTs and highlight key ongoing research areas and challenges. A few key applications that capitalize on the properties of SWCNTs are also reviewed with respect to the recent synthesis breakthroughs and ways in which synthesis science can enable advances in these applications. While the primary focus of this review is on the science framework of SWCNT growth, we draw connections to mechanisms underlying the synthesis of other 1D and 2D materials such as boron nitride nanotubes and graphene.
DIII-D physics research addresses critical challenges for the operation of ITER and the next generation of fusion energy devices. This is done through a focus on innovations to provide solutions for high performance long pulse operation, coupled with fundamental plasma physics understanding and model validation, to drive scenario development by integrating high performance core and boundary plasmas. Substantial increases in off-axis current drive efficiency from an innovative top launch system for EC power, and in pressure broadening for Alfven eigenmode control from a co-/counter-I p steerable off-axis neutral beam, all improve the prospects for optimization of future long pulse/steady state high performance tokamak operation. Fundamental studies into the modes that drive the evolution of the pedestal pressure profile and electron vs ion heat flux validate predictive models of pedestal recovery after ELMs. Understanding the physics mechanisms of ELM control and density pumpout by 3D magnetic perturbation fields leads to confident predictions for ITER and future devices. Validated modeling of high-Z shattered pellet injection for disruption mitigation, runaway electron dissipation, and techniques for disruption prediction and avoidance including machine learning, give confidence in handling disruptivity for future devices. For the non-nuclear phase of ITER, two actuators are identified to lower the L-H threshold power in hydrogen plasmas. With this physics understanding and suite of capabilities, a high poloidal beta optimized-core scenario with an internal transport barrier that projects nearly to Q = 10 in ITER at ∼8 MA was coupled to a detached divertor, and a near super H-mode optimized-pedestal scenario with co-I p beam injection was coupled to a radiative divertor. The hybrid core scenario was achieved directly, without the need for anomalous current diffusion, using off-axis current drive actuators. Also, a controller to assess proximity to stability limits and regulate β N in the ITER baseline scenario, based on plasma response to probing 3D fields, was demonstrated. Finally, innovative tokamak operation using a negative triangularity shape showed many attractive features for future pilot plant operation.
Abstract The Alpha Magnetic Spectrometer (AMS) is a precision particle physics detector on the International Space Station (ISS) conducting a unique, long-duration mission of fundamental physics research in space. The physics objectives include the precise studies of the origin of dark matter, antimatter, and cosmic rays as well as the exploration of new phenomena. Following a 16-year period of construction and testing, and a precursor flight on the Space Shuttle, AMS was installed on the ISS on May 19, 2011. In this report we present results based on 120 billion charged cosmic ray events up to multi-TeV energies. This includes the fluxes of positrons, electrons, antiprotons, protons, and nuclei. These results provide unexpected information, which cannot be explained by the current theoretical models. The accuracy and characteristics of the data, simultaneously from many different types of cosmic rays, provide unique input to the understanding of origins, acceleration, and propagation of cosmic rays.
Abstract The breast cancer risk variants identified in genome-wide association studies explain only a small fraction of the familial relative risk, and the genes responsible for these associations remain largely unknown. To identify novel risk loci and likely causal genes, we performed a transcriptome-wide association study evaluating associations of genetically predicted gene expression with breast cancer risk in 122,977 cases and 105,974 controls of European ancestry. We used data from the Genotype-Tissue Expression Project to establish genetic models to predict gene expression in breast tissue and evaluated model performance using data from The Cancer Genome Atlas. Of the 8,597 genes evaluated, significant associations were identified for 48 at a Bonferroni-corrected threshold of P < 5.82 × 10−6, including 14 genes at loci not yet reported for breast cancer. We silenced 13 genes and showed an effect for 11 on cell proliferation and/or colony-forming efficiency. Our study provides new insights into breast cancer genetics and biology.
Refractive error, measured here as mean spherical equivalent (SER), is a complex eye condition caused by both genetic and environmental factors. Individuals with strong positive or negative values of SER require spectacles or other approaches for vision correction. Common genetic risk factors have been identified by genome-wide association studies (GWAS), but a great part of the refractive error heritability is still missing. Some of this heritability may be explained by rare variants (minor allele frequency [MAF] ≤ 0.01.). We performed multiple gene-based association tests of mean Spherical Equivalent with rare variants in exome array data from the Consortium for Refractive Error and Myopia (CREAM). The dataset consisted of over 27,000 total subjects from five cohorts of Indo-European and Eastern Asian ethnicity. We identified 129 unique genes associated with refractive error, many of which were replicated in multiple cohorts. Our best novel candidates included the retina expressed PDCD6IP, the circadian rhythm gene PER3, and P4HTM, which affects eye morphology. Future work will include functional studies and validation. Identification of genes contributing to refractive error and future understanding of their function may lead to better treatment and prevention of refractive errors, which themselves are important risk factors for various blinding conditions.