Low Memory Research Articles

Pulmonary Fissure Segmentation in CT Images Using Image Filtering and Machine Learning.

Both lung lobe segmentation and lung fissure segmentation are useful in the clinical diagnosis and evaluation of lung disease. It is often of clinical interest to quantify each lobe separately because many diseases are associated with specific lobes. Fissure segmentation is important for a significant proportion of lung lobe segmentation methods, as well as for assessing fissure completeness, since there is an increasing requirement for the quantification of fissure integrity. We propose a method for the fully automatic segmentation of pulmonary fissures on lung computed tomography (CT) based on U-Net and PAN models using a Derivative of Stick (DoS) filter for data preprocessing. Model ensembling is also used to improve prediction accuracy. Our method achieved an F1 score of 0.916 for right-lung fissures and 0.933 for left-lung fissures, which are significantly higher than the standalone DoS results (0.724 and 0.666, respectively). We also performed lung lobe segmentation using fissure segmentation. The lobe segmentation algorithm shows results close to those of state-of-the-art methods, with an average Dice score of 0.989. The proposed method segments pulmonary fissures efficiently and have low memory requirements, which makes it suitable for further research in this field involving rapid experimentation.

Pre-choice midbrain fluctuations affect self-control in food choice: A functional magnetic resonance imaging (fMRI) study.

Recent studies have shown that spontaneous pre-stimulus fluctuations in brain activity affect higher-order cognitive processes, including risky decision-making, cognitive flexibility, and aesthetic judgments. However, there is currently no direct evidence to suggest that pre-choice activity influences value-based decisions that require self-control. We examined the impact of fluctuations in pre-choice activity in key regions of the reward system on self-control in food choice. In the functional magnetic resonance imaging (fMRI) scanner, 49 participants made 120 food choices that required self-control in high and low working memory load conditions. The task was designed to ensure that participants were cognitively engaged and not thinking about upcoming choices. We defined self-control success as choosing a food item that was healthier over one that was tastier. The brain regions of interest (ROIs) were the ventral tegmental area (VTA), putamen, nucleus accumbens (NAc), and caudate nucleus. For each participant and condition, we calculated the mean activity in the 3-s interval preceding the presentation of food stimuli in successful and failed self-control trials. These activities were then used as predictors of self-control success in a fixed-effects logistic regression model. The results indicate that increased pre-choice VTA activity was linked to a higher probability of self-control success in a subsequent food-choice task within the low-load condition, but not in the high-load condition. We posit that pre-choice fluctuations in VTA activity change the reference point for immediate (taste) reward evaluation, which may explain our finding. This suggests that the neural context of decisions may be a key factor influencing human behavior.

Memory-related brain potentials for visual objects in early AD show impairment and compensatory mechanisms.

Impaired episodic memory is the primary feature of early Alzheimer's disease (AD), but not all memories are equally affected. Patients with AD and amnestic Mild Cognitive Impairment (aMCI) remember pictures better than words, to a greater extent than healthy elderly. We investigated neural mechanisms for visual object recognition in 30 patients (14AD, 16 aMCI) and 36 cognitively unimpaired healthy (19 in the "preclinical" stage of AD). Event-related brain potentials (ERPs) were recorded while participants performed a visual object recognition task. Hippocampal occupancy (integrity), amyloid (florbetapir) PET, and neuropsychological measures of verbal & visual memory, executive function were also collected. A right-frontal ERP recognition effect (500-700ms post-stimulus) was seen in cognitively unimpaired participants only, and significantly correlated with memory and executive function abilities. A later right-posterior negative ERP effect (700-900ms) correlated with visual memory abilities across participants with low verbal memory ability, and may reflect a compensatory mechanism. A correlation of this retrieval-related negativity with right hippocampal occupancy (r = 0.55), implicates the hippocampus in the engagement of compensatory perceptual retrieval mechanisms. Our results suggest that early AD patients are impaired in goal-directed retrieval processing, but may engage compensatory perceptual mechanisms which rely on hippocampal function.

A New Hybrid Descent Algorithm for Large-Scale Nonconvex Optimization and Application to Some Image Restoration Problems

Conjugate gradient methods are widely used and attractive for large-scale unconstrained smooth optimization problems, with simple computation, low memory requirements, and interesting theoretical information on the features of curvature. Based on the strongly convergent property of the Dai–Yuan method and attractive numerical performance of the Hestenes–Stiefel method, a new hybrid descent conjugate gradient method is proposed in this paper. The proposed method satisfies the sufficient descent property independent of the accuracy of the line search strategies. Under the standard conditions, the trust region property and the global convergence are established, respectively. Numerical results of 61 problems with 9 large-scale dimensions and 46 ill-conditioned matrix problems reveal that the proposed method is more effective, robust, and reliable than the other methods. Additionally, the hybrid method also demonstrates reliable results for some image restoration problems.

Deep learning-based rapid prediction of temperature field and intelligent control of molten pool during directed energy deposition process

In this work, deep learning-based approaches are proposed to provide promising solutions to address the challenges in realizing intelligent manufacturing and digital twins for directed energy deposition (DED) process. Firstly, a rapid and accurate prediction of part temperature is realized by innovatively combining graph neural networks (GNNs) and recurrent neural networks (RNNs). Twenty parts with different structures are selected for demonstration. GPU parallel computing technique is adopted to accelerate the thermal finite element analysis, which is used to quickly construct the simulated graph dataset with sufficient samples. By embedding the memory optimization method into the GNN block, deeper GNNs with more trainable parameters are successfully trained with a 79.4% lower GPU memory footprint, which solves the difficulty of deeper GNNs are hard to train on large graph datasets, and the accuracy of temperature prediction on unseen DED parts is significantly improved. Secondly, for intelligent molten pool regulation, a semi-analytic temperature solution method is used to create an efficient DED environment in reinforcement learning (RL) workflows. The intelligent control of molten pool depth under complex deposition strategy is realized based on the environmental state represented by molten pool images. A tailored convolutional neural networks (CNNs) model is employed as the agent to output varying laser power and continuously interact with the dynamic environment. Compared with the traditional artificial neural network agent, the total reward scored by the CNN agent is improved by 9.7% in the zigzag deposition process, mitigating the fluctuations in the controlled molten pool depths. Moreover, CNNs are more compatible with in-situ thermal images. This work can provide theoretical and technical support for realizing real-time and even ahead-of-time temperature prediction and the corresponding feedback control during DED process.

An efficient cross-view image fusion method based on selected state space and hashing for promoting urban perception

In the field of cross-view image geolocation, traditional convolutional neural network (CNN)-based learning models generate unsatisfactory fusion performance due to their inability to model global correlations. The Transformer-based fusion methods can well compensate for the above problems, however, the Transformer has quadratic computational complexity and huge GPU memory consumption. The recent Mamba model based on the selection state space has a strong ability to model long sequences, lower GPU memory occupancy, and fewer GFLOPs. It is thus attractive and worth studying to apply Mamba to the cross-view image geolocation task. In addition, in the image-matching process (i.e., fusion of satellite/aerial and street view data.), we found that the storage occupancy of similarity measures based on floating-point features is high. Efficiently converting floating-point features into hash codes is a possible solution. In this study, we propose a cross-view image geolocation method (S6HG) based purely on Vision Mamba and hashing. S6HG fully utilizes the advantages of Vision Mamba in global information modeling and explicit location information encoding and the low storage occupancy of hash codes. Our method consists of two stages. In the first stage, we use a Siamese network based purely on vision Mamba to embed features for street view images and satellite images respectively. Our first-stage model is called S6G. In the second stage, we construct a cross-view autoencoder to further refine and compress the embedded features, and then simply map the refined features to hash codes. Comprehensive experiments show that S6G has achieved superior results on the CVACT dataset and comparable results to the most advanced methods on the CVUSA dataset. It is worth noting that other floating-point feature-based methods (4096-dimension) are 170.59 times faster than S6HG (768-bit) in storing 90,618 retrieval gallery data. Furthermore, the inference efficiency of S6G is higher than ViT-based computational methods.

Weak dangling block reordering and multi-step block compression for efficiently computing and updating PageRank solutions

The PageRank model is a powerful tool for network analysis, utilized across various disciplines such as web information retrieval, bioinformatics, community detection, and graph neural network. Computing this model requires solving a large-dimensional linear system or eigenvector problem due to the ever-increasing scale of networks. Conventional preconditioners and iterative methods for general linear systems or eigenvector problems often exhibit unsatisfactory performance for such problems, particularly as the damping factor parameter approaches 1, necessitating the development of specialized methods that exploit the specific properties of the PageRank coefficient matrix. Additionally, in practical applications, the optimal settings of the hyperparameters are generally unknown in advance, and networks often evolve over time. Consequently, recomputation of the problem is necessary following minor modifications. In this scenario, highly efficient preconditioners that significantly accelerate the iterative solution at a low memory cost are desirable. In this paper, we present two techniques that leverage the sparsity structures and numerical properties of the PageRank system, as well as a preconditioner based on the computed matrix structure. Experiments demonstrate the positive performance of the proposed methods on realistic PageRank computations.

Securing the Internet of Things with Ascon-Sign

With a Cryptographically-Relevant Quantum Computer (CRQC) estimated to be viable within the next 15 years, the development of post-quantum security is imperative. Previously secure networks may soon fall victim to these CRQCs as they will likely attack the weakest link in a network. In modern networks, these weak-links are often present in the form of Internet of Things (IoT) devices, as the resource constrains imposed by these wireless nodes leads to lowered security. We offer the first Ascon-Sign implementation for resource constrained FPGAs, which allows a wireless sensor network to verify nodes. Our design runs twice as fast as similarly-area constrained devices, and shows a 33% reduction in power per operation. We demonstrate the capability of our design by integrating it with a wireless sensor network for weather detecting. We also propose an amendment to the Ascon-Sign specification that allows for shortened processing time and lower memory requirements.

Normative data for tests of visuo-spatial, visuo-constructional skills, and visual memory for Spanish-speaking adults in the United States.

The Rey-Osterrieth Complex Figure Test (ROCFT) and the Clock Drawing Test (CDT) are commonly used in clinical practice. The ROCFT measures constructional praxis, visual perception, and visuospatial learning and memory, and the CDT assesses for visuospatial, constructional, and executive difficulties. Several neurological disorders are associated with visuospatial and visuo-constructional impairments, yet reliable normative data accounting for sociodemographic and acculturative variables are scarce for Hispanics living in the U.S. To generate normative data for the ROCFT and CDT in a Spanish-speaking adult population living in the U.S. The sample consisted of 245 cognitively healthy adults recruited from several states in the U.S. Each participant was administered the ROCFT and CDT as part of a larger cognitive battery. The ROCFT and CDT were normed using a Bayesian approach. Age, age2, education, sex, acculturation, and language proficiency were included as predictors in the analyses. ROCFT performance was associated with education and age, particularly as they interacted with Spanish language proficiency and time spent in the U.S. Education was significantly associated with recall abilities and a lower memory recall on the ROCFT. Age was found to vary depending on a person's bilingual abilities. Sex did not emerge as a predictor of performance, and it did not interact significantly with other variables. This is the first study to include acculturation and language proficiency variables in the creation of norms for the assessment of visuo-constructional abilities. This study will have a large impact on the practice of neuropsychology in the U.S.

Evaluation and Comparison of the Performance of Different Hypervisors under Various Workloads of GIADEC

This research study evaluates and compares the performance of Hyper-V, Citrix Xen, and VMware ESXi to aid Ghana Integrated Aluminium Development Corporation (GIADEC) in informed hypervisor selection. The study focuses on identifying the optimal hypervisor for GIADEC, considering factors like effectiveness and efficient resource utilization. It aims to improve resource allocation, hypervisor selection, and performance optimization within GIADEC's IT infrastructure. Using a qualitative approach and extensive literature review, the study supports informed decision-making for hypervisor selection. Through benchmarking with Passmark Performance Test Suite and BootRacer application, each hypervisor underwent rigorous testing across various performance metrics. The study assesses performance aspects including installation and booting times, CPU, memory, disk, and network performance. Results highlight distinct advantages of each hypervisor: Hyper-V for quick virtualization deployment, VMware ESXi for superior CPU, memory, and network performance. Hyper-V excels in memory write and disk sequential read/write speeds, while VMware ESXi demonstrates low memory latency and superior network bandwidth. It is recommended that GIADEC should consider VMware ESXi due to its outstanding overall performance and suitability for diverse workload requirements. This study adds to the knowledge base on hypervisor performance evaluation and supports GIADEC's decision-making process for optimal hypervisor selection.

A Long-Term Target Search Method for Unmanned Aerial Vehicles Based on Reinforcement Learning

Unmanned aerial vehicles (UAVs) are increasingly being employed in search operations. Deep reinforcement learning (DRL), owing to its robust self-learning and adaptive capabilities, has been extensively applied to drone search tasks. However, traditional DRL approaches often suffer from long training times, especially in long-term search missions for UAVs, where the interaction cycles between the agent and the environment are extended. This paper addresses this critical issue by introducing a novel method—temporally asynchronous grouped environment reinforcement learning (TAGRL). Our key innovation lies in recognizing that as the number of training environments increases, agents can learn knowledge from discontinuous trajectories. This insight leads to the design of grouped environments, allowing agents to explore only a limited number of steps within each interaction cycle rather than completing full sequences. Consequently, TAGRL demonstrates faster learning speeds and lower memory consumption compared to existing parallel environment learning methods. The results indicate that this framework enhances the efficiency of UAV search tasks, paving the way for more scalable and effective applications of RL in complex scenarios.

Utilizing Inherent Bias for Memory Efficient Continual Learning: A Simple and Robust Baseline

Learning from continuously evolving data is critical in real-world applications. This type of learning, known as Continual Learning (CL), aims to assimilate new information without compromising performance on prior knowledge. However, learning new information leads to a bias in the network towards recent observations, resulting in a phenomenon known as catastrophic forgetting. The complexity increases in Online Continual Learning (OCL) scenarios where models are allowed only a single pass over data. Existing OCL approaches that rely on replaying exemplar sets are not only memory-intensive when it comes to large-scale datasets but also raise security concerns. While recent dynamic network models address memory concerns, they often present computationally demanding, over-parameterized solutions with limited generalizability. To address this longstanding problem, we propose a novel OCL approach termed “Bias Robust online Continual Learning (BRCL).” BRCL retains all intermediate models generated. These models inherently exhibit a preference for recently learned classes. To leverage this property for enhanced performance, we devise a strategy we describe as ‘utilizing bias to counteract bias.’ This method involves the development of an Inference function that capitalizes on the inherent biases of each model towards the recent tasks. Furthermore, we integrate a model consolidation technique that aligns the first layers of these models, particularly focusing on similar feature representations. This process effectively reduces the memory requirement, ensuring a low memory footprint. Despite the simplicity of the methodology to guarantee expandability to various frameworks, extensive experiments reveal a notable performance edge over leading methods on key benchmarks, getting continual learning closer to matching offline training. (Source code will be made publicly available upon the publication of this paper.)

Physics-Informed Holomorphic Neural Networks (PIHNNs): Solving 2D linear elasticity problems

We propose physics-informed holomorphic neural networks (PIHNNs) as a method to solve boundary value problems where the solution can be represented via holomorphic functions. Specifically, we consider the case of plane linear elasticity and, by leveraging the Kolosov–Muskhelishvili representation of the solution in terms of holomorphic potentials, we train a complex-valued neural network to fulfill stress and displacement boundary conditions while automatically satisfying the governing equations. This is achieved by designing the network to return only approximations that inherently satisfy the Cauchy-Riemann conditions through specific choices of layers and activation functions. To ensure generality, we provide a universal approximation theorem guaranteeing that, under basic assumptions, the proposed holomorphic neural networks can approximate any holomorphic function. Furthermore, we suggest a new tailored weight initialization technique to mitigate the issue of vanishing/exploding gradients. Compared to the standard PINN approach, noteworthy benefits of the proposed method for the linear elasticity problem include a more efficient training, as evaluations are needed solely on the boundary of the domain, lower memory requirements, due to the reduced number of training points, and C∞ regularity of the learned solution. Several benchmark examples are used to verify the correctness of the obtained PIHNN approximations, the substantial benefits over traditional PINNs, and the possibility to deal with non-trivial, multiply-connected geometries via a domain-decomposition strategy.

Reminders eliminate age-related declines in prospective memory.

Prospective memory (PM)-the process of establishing intentions for future action and remembering to fulfill these intentions at the appropriate time-is necessary for many instrumental activities of daily living and for maintaining functional independence with increased age. Offloading PM demands onto the environment, such as setting a reminder alarm to take medication, offers an easy and effective way to mitigate age-related PM declines. However, a lack of basic knowledge about the cognitive and metacognitive processes that drive offloading decisions presents barriers to successful implementation. The present study addresses these issues by examining age differences in PM for offloaded (i.e., with reminders) and nonoffloaded (i.e., without reminders) intentions under low and high memory demands. With highly specific intentions that can be retrieved via bottom-up processes (Experiment 1), there were no age differences in PM without reminders, and younger and older adults equally benefitted from reminders under high memory load. With nonspecific intentions that require top-down attention for retrieval (Experiment 2), older adults had worse PM under high load without reminders. Critically, this age difference was eliminated with the use of reminders, likely due to increased reminder checking for older adults under high load. These findings suggest that offloading can circumvent cognitive capacity limitations and minimize computational effort to improve intention fulfillment in older adults. The theoretical and applied ramifications of these findings are discussed. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

T Cell Peptide Prediction, Immune Response, and Host-Pathogen Relationship in Vaccinated and Recovered from Mild COVID-19 Subjects.

COVID-19 remains a significant threat, particularly to vulnerable populations. The emergence of new variants necessitates the development of treatments and vaccines that induce both humoral and cellular immunity. This study aimed to identify potentially immunogenic SARS-CoV-2 peptides and to explore the intricate host-pathogen interactions involving peripheral immune responses, memory profiles, and various demographic, clinical, and lifestyle factors. Using in silico and experimental methods, we identified several CD8-restricted SARS-CoV-2 peptides that are either poorly studied or have previously unreported immunogenicity: fifteen from the Spike and three each from non-structural proteins Nsp1-2-3-16. A Spike peptide, LA-9, demonstrated a 57% response rate in ELISpot assays using PBMCs from 14 HLA-A*02:01 positive, vaccinated, and mild-COVID-19 recovered subjects, indicating its potential for diagnostics, research, and multi-epitope vaccine platforms. We also found that younger individuals, with fewer vaccine doses and longer intervals since infection, showed lower anti-Spike (ELISA) and anti-Wuhan neutralizing antibodies (pseudovirus assay), higher naïve T cells, and lower central memory, effector memory, and CD4hiCD8low T cells (flow cytometry) compared to older subjects. In our cohort, a higher prevalence of Vδ2-γδ and DN T cells, and fewer naïve CD8 T cells, seemed to correlate with strong cellular and lower anti-NP antibody responses and to associate with Omicron infection, absence of confusional state, and habitual sporting activity.

A FinFET based Low Leakage 10T Static Random Access Memory cell

A FinFET based Low Leakage 10T Static Random Access Memory cell

M-adapter: Multi-level image-to-video adaptation for video action recognition

With the growing size of visual foundation models, training video models from scratch has become costly and challenging. Recent attempts focus on transferring frozen pre-trained Image Models (PIMs) to video fields by tuning inserted learnable parameters such as adapters and prompts. However, these methods require saving PIM activations for gradient calculations, leading to limited savings of GPU memory. In this paper, we propose a novel parallel branch that adapts the multi-level outputs of the frozen PIM for action recognition. It avoids passing gradients through the PIMs, thus naturally owning much lower GPU memory footprints. The proposed adaptation branch consists of hierarchically combined multi-level output adapters (M-adapters), comprising a fusion module and a temporal module. This design digests the existing discrepancies between the pre-training task and the target task with lower training costs. We show that when using larger models or on scenarios with higher demands for temporal modelling, the proposed method performs better than those with the full-parameter tuning manner. Finally, despite only tuning fewer parameters, our method achieves superior or comparable performance against current state-of-the-art methods.

MiniSNV: accurate and fast single nucleotide variant calling from nanopore sequencing data.

Nanopore sequence technology has demonstrated a longer read length and enabled to potentially address the limitations of short-read sequencing including long-range haplotype phasing and accurate variant calling. However, there is still room for improvement in terms of the performance of single nucleotide variant (SNV) identification and computing resource usage for the state-of-the-art approaches. In this work, we introduce miniSNV, a lightweight SNV calling algorithm that simultaneously achieves high performance and yield. miniSNV utilizes known common variants in populations as variation backgrounds and leverages read pileup, read-based phasing, and consensus generation to identify and genotype SNVs for Oxford Nanopore Technologies (ONT) long reads. Benchmarks on real and simulated ONT data under various error profiles demonstrate that miniSNV has superior sensitivity and comparable accuracy on SNV detection and runs faster with outstanding scalability and lower memory than most state-of-the-art variant callers. miniSNV is available from https://github.com/CuiMiao-HIT/miniSNV.

Monitoring Cognitive Functions During Deep Brain Stimulation Interventions by Real Time Neuropsychological Testing.

We monitored cognition in 14 Parkinson's disease (PD) patients during deep brain stimulation (DBS) surgery when the electrode was positioned at the target subthalamic nucleus (STN) (i.e., the STN motor area). We present the DBS-real-time neuropsychological testing (DBS-RTNT) protocol and our preliminary experience with it; we also compared the intraoperative patient performance with the baseline data. Compared with the baseline data, patients undergoing DBS-RTNT in the target area demonstrated a significantly decreased performance on some tasks belonging to the memory and executive function domains. Patients undergoing right hemisphere DBS-RTNT had significantly lower short-term memory and sequencing scores than did patients undergoing left hemisphere DBS-RTNT. PD patient cognitive performance should be monitored during DBS surgery, as STN-DBS may induce changes. These preliminary data contribute to improving our understanding of the anatomo-functional topography of the STN during DBS surgery, which will enable the identification of the best site for producing positive motor effects without causing negative cognitive and/or emotional changes in individual patients in the future. In principle, medications (i.e., patients who underwent surgery in a levodopa-off state) could have influenced our results; therefore, future studies are needed to address the possible confounding effects of levodopa use.

Association between subcortical nuclei volume changes and cognition in preschool-aged children with tetralogy of Fallot after corrective surgery: a cross-sectional study

BackgroundNeurocognitive disorders frequently occur in patients with cyanotic congenital heart disease (CCHD) because of the hemodynamic abnormalities induced by preoperative cardiac structural changes. We aimed to evaluate subcortical nuclei volume changes and cognition in postoperative tetralogy of Fallot (TOF) children, and analyze their relationship with preoperative cardiac structural changes.MethodsThis case-control study involved thirty-six children with repaired TOF and twenty-nine healthy controls (HCs). We utilized three-dimensional (3D) T1-weighted high-resolution structural images alongside the Wechsler Preschool and Primary Scale of Intelligence-Fourth Edition (WPPSI-IV) to evaluate the cognitive differences between the TOF and HC group.ResultsWe observed notable differences in subcortical nuclei volume between the TOF and HC group, specifically in the left amygdala nucleus (LAM, TOF: 1292.60 ± 155.57; HC: 1436.27 ± 140.62, p < 0.001), left thalamus proper nucleus (LTHA, TOF: 6771.54 ± 666.03; HC: 7435.36 ± 532.84, p < 0.001), and right thalamus proper nucleus (RTHA, TOF: 6514.61 ± 715.23; HC: 7162.94 ± 554.60, p < 0.001). Furthermore, a diminished integrity of LAM ( β:-19.828, 95% CI: -36.462, -3.193), which showed an inverse relationship with the size of the preoperative ventricular septal defect (VSD), correlated with lower working memory indices in children with TOF.ConclusionsOur findings indicate that subcortical nuclei structural injuries possibly potentially stemming from cardiac anatomical abnormalities, are associated with impaired working memory in preschool-aged children with TOF. The LAM in particular may serve as a potential biomarker for neurocognitive deficits in TOF, offering predictive value for future neurodevelopmental outcomes, and shedding light on the neurophysiological mechanisms of these cognitive impairments.