Variety Of Architectures Research Articles

Exploring the Impact of Architectural Variations in ResNet on CIFAR-100 Performance: An Investigation on Fully Connected Layers, Residual Blocks, and Kernel Size

ResNet is one of the leading neutral networks that has been widely applied in image classification. This study built a simple baseline network based on the concept of ResNet and then examines how variations in ResNet’s architecture affect its model performance to provide insights for optimizing network design. Firstly, this study investigates the number of fully connected layers, the results show that by reducing the number of fully connected layers significantly decreases the total number of trainable parameters, which in turn reduces the training time. However, this reduction does not lead to a noticeable improvement in the accuracy after convergence. In addition, increasing the number of fully connected layers not only greatly increases the training time but also leads to overfitting on the CIFAR-100 dataset, slightly reducing the training performance. Secondly, this study also analyzes the influence of reducing number of residual basic block. Analysis suggests that reducing the use of Residual Blocks has significantly negatively impacted both accuracy and training time. This may be because the use of Residual Blocks positively affects the network’s ability to learn features from the CIFAR-100 dataset. Finally, this study explores the effect of bigger kernel size of convolution layer in Residual Basic Block. The outcome demonstrates increasing the kernel size in the Residual Blocks significantly improves both training time and accuracy. Additionally, it was observed that in this variant experiment, a definitive convergence has not yet been clearly established, leaving the possibility that accuracy might continue to improve with more training epochs.

On a Simplified Approach to Achieve Parallel Performance and Portability Across CPU and GPU Architectures

This paper presents software advances to easily exploit computer architectures consisting of a multi-core CPU and CPU+GPU to accelerate diverse types of high-performance computing (HPC) applications using a single code implementation. The paper describes and demonstrates the performance of the open-source C++ matrix and array (MATAR) library that uniquely offers: (1) a straightforward syntax for programming productivity, (2) usable data structures for data-oriented programming (DOP) for performance, and (3) a simple interface to the open-source C++ Kokkos library for portability and memory management across CPUs and GPUs. The portability across architectures with a single code implementation is achieved by automatically switching between diverse fine-grained parallelism backends (e.g., CUDA, HIP, OpenMP, pthreads, etc.) at compile time. The MATAR library solves many longstanding challenges associated with easily writing software that can run in parallel on any computer architecture. This work benefits projects seeking to write new C++ codes while also addressing the challenges of quickly making existing Fortran codes performant and portable over modern computer architectures with minimal syntactical changes from Fortran to C++. We demonstrate the feasibility of readily writing new C++ codes and modernizing existing codes with MATAR to be performant, parallel, and portable across diverse computer architectures.

Genomic prediction of cereal crop architectural traits using models informed by gene regulatory circuitries from maize.

Plant architecture is a major determinant of planting density, which enhances productivity potential for crops per unit area. Genomic prediction is well positioned to expedite genetic gain of plant architectural traits since they are typically highly heritable. Additionally, the adaptation of genomic prediction models to query predictive abilities of markers tagging certain genomic regions could shed light on the genetic architecture of these traits. Here, we leveraged transcriptional networks from a prior study that contextually described developmental progression during tassel and leaf organogenesis in maize (Zea mays) to inform genomic prediction models for architectural traits. Since these developmental processes underlie tassel branching and leaf angle, 2 important agronomic architectural traits, we tested whether genes prioritized from these networks quantitatively contribute to the genetic architecture of these traits. We used genomic prediction models to evaluate the ability of markers in the vicinity of prioritized network genes to predict breeding values of tassel branching and leaf angle traits for 2 diversity panels in maize and diversity panels from sorghum (Sorghum bicolor) and rice (Oryza sativa). Predictive abilities of markers near these prioritized network genes were similar to those using whole-genome marker sets. Notably, markers near highly connected transcription factors from core network motifs in maize yielded predictive abilities that were significantly greater than expected by chance in not only maize but also closely related sorghum. We expect that these highly connected regulators are key drivers of architectural variation that are conserved across closely related cereal crop species.

Computational Study of SmI2-Catalyzed Intermolecular Couplings of Cyclopropyl Ketones: Links between the Structure and Reactivity.

SmI2-catalyzed intermolecular coupling reactions of cyclopropyl ketones with alkenes or alkynes offer an efficient strategy for furnishing diverse five-membered ring-containing molecular architectures. This study presents a systematic computational investigation to reveal the structure-reactivity relationships in these reactions. The reactivity of aryl cyclopropyl ketones is enhanced by the stabilized ketyl radical and cyclopropyl fragmentation, arising from the conjugation effect of the aryl ring, despite an obstacle emerging from the gauche styrene intermediate that elevates the energy barrier for radical trapping. By contrast, alkyl cyclopropyl ketones lack conjugation and exhibit high barriers for reduction and fragmentation but undergo facile radical trapping due to the minimal steric hindrance. Interestingly, ortho-substituted phenyl cyclopropyl ketones exhibit superior reactivity due to a balance between the moderate conjugation, promoting cyclopropyl fragmentation, and the pretwisted nature of the ortho-substituted phenyl that circumvents the hindrance posed by the gauche intermediate and facilitates the radical trapping. The markedly enhanced reactivity of bicyclo[1.1.0]butyl (BCB) ketones arises from facile fragmentation of the strained BCB motif. Bicyclo[2.1.0]pentyl (BCP) ketones, less strained than BCB ketones, are computationally verified to undergo efficient couplings with various partners, and this can be attributed to their stable fragmentation intermediates that facilitate radical trapping. Our findings provide insights that can aid in designing related reactions.

Cell wall remodeling confers plant architecture with distinct wall structure in Nelumbo nucifera.

Lotus (Nelumbo nucifera G.) is a perennial aquatic horticultural plant with diverse architectures. Distinct plant architecture (PA) has certain attractive and practical qualities, but its genetic morphogenesis in lotus remains elusive. In this study, we employ genome-wide association analysis (GWAS) for the seven traits of petiole length (PLL), leaf length (LL), leaf width (LW), peduncle length (PLF), flower diameter (FD), petal length (PeL), and petal width (PeW) in 301 lotus accessions. A total of 90 loci are identified to associate with these traits across 4 years of trials. Meanwhile, we perform RNA sequencing (RNA-seq) to analyze the differential expression of the gene (DEG) transcripts between large and small PA (LPA and SPA) of lotus stems (peduncles and petioles). As a result, eight key candidate genes are identified that are all primarily involved in plant cell wall remodeling significantly associated with PA traits by integrating the results of DEGs and GWAS. To verify this result, we compare the cell wall compositions and structures of LPA versus SPA in representative lotus germplasms. Intriguingly, compared with the SPA lotus, the LPA varieties have higher content of cellulose and hemicellulose, but less filling substrates of pectin and lignin. Additionally, we verified longer cellulose chains and higher cellulose crystallinity with less interference in LPA varieties. Taken together, our study illustrates how plant cell wall remodeling affects PA in lotus, shedding light on the genetic architecture of this significant ornamental trait and offering a priceless genetic resource for future genomic-enabled breeding.

Form Equals Function: Influence of Coacervate Architecture on Drug Delivery Applications.

Complex coacervates, formed through electrostatic interactions between oppositely charged polymers, present a versatile platform for drug delivery, providing rapid assembly, selective encapsulation, and responsiveness to environmental stimuli. The architecture and properties of coacervates can be tuned by controlling structural and environmental design factors, which significantly impact the stability and delivery efficiency of the drugs. While environmental design factors such as salt, pH, and temperature play a crucial role in coacervate formation, structural design factors such as polymer concentration, polymer structure, mixing ratio, and chain length serve as the core framework that shapes coacervate architecture. These elements modulate the phase behavior and material properties of coacervates, allowing for a highly tunable system. In this review, we primarily analyze how these structural design factors contribute to the formation of diverse coacervate architecture, ranging from bulk coacervates to polyion complex micelles, vesicles, and cross-linked gels, though environmental design factors are considered. We then examine the effectiveness of these architectures in enhancing the delivery and efficacy of drugs across various administration routes, such as noninvasive (e.g., oral and transdermal) and invasive delivery. This review aims to provide foundational insights into the design of advanced drug delivery systems by examining how the origin and chemical structure of polymers influence coacervate architecture, which in turn defines their material properties. We then explore how the architecture can be tailored to optimize drug delivery for specific administration routes. This approach leverages the intrinsic properties derived from the coacervate architecture to enable targeted, controlled, and efficient drug release, ultimately enhancing therapeutic outcomes in precision medicine.

Allocyclic and autocyclic controls on facies distribution in interdistributary bayfill deposits of the Campanian Neslen Formation, Eastern Book Cliffs, Utah, U.S.A.

ABSTRACT The Campanian Neslen Formation, in the eastern part of the Book Cliffs, is divided into three stratigraphic intervals. The middle Neslen interval consists of paralic bayfill deposits that average 23 m in thickness. This article describes the lateral and vertical variability within the bayfills deposited in this middle Neslen interval. Comprehensive mapping along 27 km of outcrops, including high-resolution description of 62 sedimentary logs and incorporation of wireline log data from 86 wells, has led to a detailed understanding of the processes related to infilling of interdistributary bays. Six vertically stacked heterolithic to sand-prone bayfill units are identified. A distinction is made between bayfill units and bayfill successions, based on their bounding autocyclic or allocyclic flooding surfaces. Units are bounded by high-frequency flooding surfaces, and show large lateral variations in thickness, facies associations, and lithology. These bayfill units are laterally subdivided into a varying number of bayfill successions, separated by interdistributary channel systems or associated bayhead-delta facies associations. Lateral variation in facies architecture, salinity, and depositional energy levels are mapped in each bayfill succession. This variation is attributed to several controlling factors, including proximity to sediment entry points, water depth, the sequence stratigraphic position of the bayfill succession, and the compaction of underlying sandbodies. The detailed facies patterns and multitude of tributary causes described and discussed here are applicable to an overall understanding of facies architecture in paralic environments.

The Temple as a Tangible Heritage of India: A Living Testament to Culture, Spirituality, and Architecture

The temples of India stand as a remarkable testimony to the nation’s spiritual, cultural, and architectural heritage. This paper delves into the multifaceted role of temples as tangible heritage, tracing their evolution from the early Gupta period to the architectural marvels of the Chola and Chandela dynasties. It explores the architectural diversity across regions, from the Nagara and Dravida styles to the unique Kalinga and Vesara forms. Temples are not just religious spaces but have historically served as cultural, social, and economic centers, nurturing art, education, and local economies. Their architectural brilliance, rich symbolism, and sacred geometry embody profound spiritual philosophies. Moreover, temples reflect the political and social history of India, with kings commissioning grand edifices to solidify their rule and connect with the divine. The preservation of these temples, many of which are threatened by neglect, environmental factors, and human activity, is crucial. Governmental and international efforts, including UNESCO designations, play a significant role in conserving these living monuments. As India modernizes, safeguarding its temple heritage is essential to maintaining the continuity of its cultural and spiritual identity for future generations.

Ranking the Transferability of Adversarial Examples

Adversarial transferability in blackbox scenarios presents a unique challenge: while attackers can employ surrogate models to craft adversarial examples, they lack assurance on whether these examples will successfully compromise the target model. Until now, the prevalent method to ascertain success has been trial and error—testing crafted samples directly on the victim model. This approach, however, risks detection with every attempt, forcing attackers to either perfect their first try or face exposure. Our article introduces a ranking strategy that refines the transfer attack process, enabling the attacker to estimate the likelihood of success without repeated trials on the victim’s system. By leveraging a set of diverse surrogate models, our method can predict transferability of adversarial examples. This strategy can be used to either select the best sample to use in an attack or the best perturbation to apply to a specific sample. Using our strategy, we were able to raise the transferability of adversarial examples from a mere 20%—akin to random selection—up to near upper-bound levels, with some scenarios even witnessing a 100% success rate. This substantial improvement not only sheds light on the shared susceptibilities across diverse architectures but also demonstrates that attackers can forego the detectable trial-and-error tactics raising increasing the threat of surrogate-based attacks.

AI-Driven Hardware Testing: Overcoming the Challenges of Modern Hardware Architecture and Power Management

With the rapid development of hardware technology, the diversity of architectures, the complexity of hardware modules, and the growing demand for real-time capabilities, enhanced power management, adaptability, and flexibility all present new challenges. These challenges include managing the complexity of hardware architecture, performance, and energy efficiency, which traditional testing methods struggle to address effectively. To solve these problems, this paper proposes a new standard for integrating Artificial intelligence with hardware test systems. The system can automatically identify key performance indicators according to the hardware characteristics, dynamically adjust the test strategy, optimize the test process, improve efficiency, and ultimately shorten the time to market.

ZeroD-fender: A Resource-aware IoT Malware Detection Engine via Fine-grained Side-channel Analysis

In early 2023, cyberattacks experienced a significant rise due to unknown (zero-day) malware targeting Internet of Things (IoT) devices. To tackle the challenge of zero-day detection within a highly resource-constrained IoT environment, we propose a novel design that utilizes fine-grained power side-channel analysis with deep learning techniques. Our approach introduces an innovative concept called multiscale feature extraction to identify the most representative malware features across diverse architectures, thereby enhancing deep learning based detection performance against zero-day malware. Specifically, we employ a fine-grained power side-channel analysis of more than 120,000 honeypot-collected malware files across a hierarchy of commands , functions , and modules to identify the unique zero-day malware behaviors. With these identified features to train our model, ZeroD-fender’s performance in detecting zero-day malware has significantly improved. In pursuit of on-device detection, we present a resource-aware online inference customization framework. This framework features our lightweight network, ThingNetV2, which uses specialized 1-D depthwise separable convolution paired with h-swish activation, leading to significant resource savings. By applying the fine-grained power analysis, ZeroD-fender demonstrates a detection rate of 95.88% across various architecture zero-day malware, achieving detection speeds ranging between 16.083 ms and 23.961 ms , depending on the specific scenario.

Morpho-Colorimetric Diversity and Genome-Wide Association Study of Wheat Spike Architecture Based on Image Processing Under Well-Watered and Rain-Fed Conditions

Morpho-Colorimetric Diversity and Genome-Wide Association Study of Wheat Spike Architecture Based on Image Processing Under Well-Watered and Rain-Fed Conditions

Equimolar Cross-Coupling Using Reactive Coiled Coils for Covalent Protein Assemblies.

Biocompatible cross-coupling reactions enable the efficient covalent attachment of large biomolecules at near-stoichiometric ratios, ensuring the stability and integrity of the resulting products. We present an affinity-based peptide platform utilizing coiled coils containing reactive side chains for proximity-driven protein cross-coupling in the presence of a cross-linking agent. This platform supports both chemical synthesis and recombinant expression, using canonical amino acids to generate reactive affinity tags. Employing the E3/R3 coiled coil pair as a scaffold, we design four complementary coils with cysteine residues as cross-linking sites, achieving >90% conversion to covalent heterodimeric coupling products using 3,4-dibromomaleimide. Equimolar mixtures of proteins with reactive coils at their termini yield near-quantitative heterodimeric cross-coupling products. The strategic selection of complementary coiled coil pairs and cross-linking agents enables orthogonal assembly of macromolecules with diverse architectures. This method offers a versatile approach for creating covalent fusion proteins, enhancing their stability and functionality for applications in chemical biology, biotechnology, and medicine.

Exploring Anatomical Variations in the Bony Architecture of the Greater Palatine Canal in Dry Bones

Exploring Anatomical Variations in the Bony Architecture of the Greater Palatine Canal in Dry Bones

Evolution of Thylakoid Structural Diversity.

Oxygenic photosynthesis evolved billions of years ago, becoming Earth's main source of biologically available carbon and atmospheric oxygen. Since then, phototrophic organisms have diversified from prokaryotic cyanobacteria into several distinct clades of eukaryotic algae and plants through endosymbiosis events. This diversity can be seen in the thylakoid membranes, complex networks of lipids, proteins, and pigments that perform the light-dependent reactions of photosynthesis. In this review, we highlight the structural diversity of thylakoids, following the evolutionary history of phototrophic species. We begin with a molecular inventory of different thylakoid components and then illustrate how these building blocks are integrated to form membrane networks with diverse architectures. We conclude with an outlook on understanding how thylakoids remodel their architecture and molecular organization during dynamic processes such as biogenesis, repair, and environmental adaptation.

Anti-inflammatory bicyclic polyprenylated acylphloroglucinols with diverse architectures including an unprecedented 6/6/6 tricyclic core from Garcinia yunnanensis

Garciyunnanol A (1), an unprecedented 1,2-seco-bicyclic polyprenylated acylphloroglucinol (BPAP) possessing a unique 6/6/6 tricyclic core, was characterized from Garcinia yunnanensis together with 16 BPAPs, including eight new compounds (garciyunnanols B–I, 2–9). Biogenetically, the bicyclo[3.3.1]nonane-2,4,9-trione moiety of 12 reconstructed the bicyclic δ-lactone core of 2 through Norrish type Ⅰ cleavage and cyclization, followed by a cyclization of two side chains to form an intriguing 6/6/6 tricyclic core of 1. Their structures were elucidated through analysis of spectroscopic data, calculation and comparison of ECD spectra. Bioactivity evaluation manifested that compounds 1, 2, 5, 6 and 14 demonstrated superior inhibition of NO production compared to the positive control dexamethasone. Notably, compound 5 exhibited a dose-dependent inhibitory effect on NO production, with an IC50 value of 0.25 ± 0.87 µM. Furthermore, experiments involving ELISA, Western blotting, and immunofluorescence staining revealed that 5 effectively reduced the secretion of interleukin-1β in LPS plus nigericin-stimulated THP-1 macrophages by inhibiting the activation of the NLRP3 inflammasome.

Prototype-Based Semantic Segmentation.

Deep learning based semantic segmentation solutions have yielded compelling results over the preceding decade. They encompass diverse network architectures (FCN based or attention based), along with various mask decoding schemes (parametric softmax based or pixel-query based). Despite the divergence, they can be grouped within a unified framework by interpreting the softmax weights or query vectors as learnable class prototypes. In light of this prototype view, we reveal inherent limitations within the parametric segmentation regime, and accordingly develop a nonparametric alternative based on non-learnable prototypes. In contrast to previous approaches that entail the learning of a single weight/query vector per class in a fully parametric manner, our approach represents each class as a set of non-learnable prototypes, relying solely upon the mean features of training pixels within that class. The pixel-wise prediction is thus achieved by nonparametric nearest prototype retrieving. This allows our model to directly shape the pixel embedding space by optimizing the arrangement between embedded pixels and anchored prototypes. It is able to accommodate an arbitrary number of classes with a constant number of learnable parameters. Through empirical evaluation with FCN based and Transformer based segmentation models (i.e., HRNet, Swin, SegFormer, Mask2Former) and backbones (i.e., ResNet, HRNet, Swin, MiT), our nonparametric framework shows superior performance on standard segmentation datasets (i.e., ADE20 K, Cityscapes, COCO-Stuff), as well as in large-vocabulary semantic segmentation scenarios. We expect that this study will provoke a rethink of the current de facto semantic segmentation model design.

HKNAS: Classification of Hyperspectral Imagery Based on Hyper Kernel Neural Architecture Search.

Recent neural architecture search (NAS)-based approaches have made great progress in the hyperspectral image (HSI) classification tasks. However, the architectures are usually optimized independently of the network weights, increasing searching time, and restricting model performances. To tackle these issues, in this article, different from previous methods that extra define structural parameters, we propose to directly generate structural parameters by utilizing the specifically designed hyper kernels, ingeniously converting the original complex dual optimization problem into easily implemented one-tier optimizations, and greatly shrinking searching costs. Then, we develop a hierarchical multimodule search space whose candidate operations only contain convolutions, and these operations can be integrated into unified kernels. Using the above searching strategy and searching space, we obtain three kinds of networks to separately conduct pixel-level or image-level classifications with 1-D or 3-D convolutions. In addition, by combining the proposed hyper kernel searching scheme with the 3-D convolution decomposition mechanism, we obtain diverse architectures to simulate 3-D convolutions, greatly improving network flexibilities. A series of quantitative and qualitative experiments on six public datasets demonstrate that the proposed methods achieve state-of-the-art results compared with other advanced NAS-based HSI classification approaches.

Innovative Poly(lactic Acid) Blends: Exploring the Impact of the Diverse Chemical Architectures from Itaconic Acid

Environment-friendly polymer blends of poly(lactic acid) (PLA) and itaconic acid (IA), poly(itaconic acid) (PIA), poly(itaconic acid)-co-poly(methyl itaconate) (Cop-IA), and net-poly(itaconic acid)-ν-triethylene glycol dimethacrylate (Net-IA) were performed via melt blending. The compositions studied were 0.1, 1, 3, and 10 wt% of the diverse chemical architectures. The research aims to study and understand the effect of IA and its different architectures on the mechanical, rheological, and thermal properties of PLA. The PLA/IA, PLA/PIA, PLA/Cop-IA, and PLA/Net-IA blends were characterized by dynamic mechanical thermal analysis, rotational rheometer (RR), thermogravimetric analysis, differential scanning calorimetry, X-ray diffraction, and scanning electron microscopy. The complex viscosity, storage module, and loss module for the RR properties were observed in the following order: PLA/Cop-IA, PLA/Net-IA, and PLA/PIA > PLA > PLA/IA. Thermal stability improved with increasing concentrations of Cop-IA and Net-IA. In the same way, the mechanical properties were enhanced. In addition, the micrographs illustrated the formation of fibrillar structures for all blends. The crystallinity degree displayed higher values for the blends that contain Net-IA > Cop-IA than IA > PIA. Therefore, IA and its architectures can influence these studied properties, which have potential applications in disposable food packing.

Investigation of colonial-era architecture style on modern architecture design style in Aden city

This study investigated the lasting influence of British colonial architecture (129 years) on contemporary building design in Aden, Yemen. The initial focus was on the perception of British Neo-Classical architecture as a symbol of superiority, overshadowing Aden's rich architectural history. This research explored Aden's architectural transformation from pre-colonial to colonial periods, analyzing the impact of colonial-era building facades on modern design practices. Through observation and interviews with built environment professionals, the study examined the broader social, political, and economic effects of British colonialism. The findings reveal a significant influence of British colonial architecture on the design of building facades in Aden. This influence is evident in structures like the repurposed church on Crater Mountain, which now serves as a government building. The influx of British administrators, soldiers, and workers from diverse backgrounds (Indian, Somali, Jewish) during the colonial period contributed to Aden's architectural diversity by constructing churches, temples, and government buildings. Unfortunately, the ongoing war in Yemen has damaged or destroyed many of these structures. This study also found that classical architectural elements from British colonial buildings like the Military Museum in Crater, the Heritage Museum, and the Tax Office continue to influence modern design. These elements and the miniature Big Ben clock in Al-Tawahi stand as prominent symbols of British influence in Aden. However, the Yemeni government's restoration efforts for these colonial-era buildings have been slow and largely ineffective. The research further explored the enduring influence of Neo-classical architectural ideals in Aden. The study highlights the formal guidelines, specific modules, and proportions used in designing arches and columns, particularly in residential buildings. These elements contribute to the aesthetic qualities of these structures. While the British incorporated classical elements during their rule, architectural advancements and new materials have led to the incorporation of novel elements in contemporary design.