Memory Mechanisms Research Articles

A Cross-Client Coordinator in Federated Learning Framework for Conquering Heterogeneity.

Federated learning, as a privacy-preserving learning paradigm, restricts the access to data of each local client, for protecting the privacy of the parties. However, in the case of heterogeneous data settings, the different data distributions among clients usually lead to the divergence of learning targets, which is an essential challenge for federated learning. In this article, we propose a federated learning framework with a unified coding space, called FedUCS, for learning cross-client uniform coding rules to solve the problem of divergent targets among multiple clients due to heterogeneous data. A cross-client coordinator co-trained by multiple clients is used as a criterion of the coding space to supervise all clients coding to a uniform space, which is the significant contribution of this article. Furthermore, in order to appropriately retain historical information and avoid forgetting previous knowledge, a partial memory mechanism is applied. Moreover, in order to further enhance the distinguishability of the unified encoding space, supervised contrastive learning is used to avoid the intersection of the encoding spaces belonging to different categories. A series of experiments are performed to verify the effectiveness of the proposed method in a federated learning setting with heterogeneous data.

Muscle memory in humans: evidence for myonuclear permanence and long-term transcriptional regulation after strength training.

The objective of this work was to investigate myonuclear permanence and transcriptional regulation as mechanisms for cellular muscle memory after strength training in humans. Twelve untrained men and women performed 10weeks of unilateral elbow-flexor strength training followed by 16weeks of de-training. Thereafter, 10weeks' re-training was conducted with both arms: the previously trained arm and the contralateral untrained control arm. Muscle biopsies were taken from the trained arm before and after both training periods and from the control arm before and after re-training. Muscle biopsies were analysed for fibre cross-sectional area (fCSA), myonuclei and global transcriptomics (RNA sequencing). During the first training period, myonuclei increased in type 1 (13 ± 17%) and type 2 (33 ± 23%) fibres together with a 30 ± 43% non-significant increase in mixed fibre fCSA (P=0.069). Following de-training, fCSA decreased in both fibre types, whereas myonuclei were maintained, resulting in 33% higher myonuclear number in previously trained vs. control muscle in type 2 fibres. Furthermore, in the previously trained muscle, three differentially expressed genes (DEGs; EGR1, MYL5 and COL1A1) were observed. Following re-training, the previously trained muscle showed larger type 2 fCSA compared to the control (P=0.035). However, delta change in type 2 fCSA was not different between muscles. Gene expression was more dramatically changed in the control arm (1338 DEGs) than in the previously trained arm (822 DEGs). The sustained higher number of myonuclei in the previously trained muscle confirms myonuclear accretion and permanence in humans. Nevertheless, because of the unclear effect on the subsequent hypertrophy with re-training, the physiological benefit remains to be determined. KEY POINTS: Muscle memory is a cellular mechanism that describes the capacity of skeletal muscle fibres to respond differently to training stimuli if the stimuli have been previously encountered. This study overcomes past methodological limitations related to the choice of muscles and analytical procedures. We show that myonuclear number is increased after strength training and maintained during de-training. Increased myonuclear number and differentially expressed genes related to muscle performance and development in the previously trained muscle did not translate into a clearly superior responses during re-training. Because of the unclear effect on the subsequent hypertrophy and muscle strength gain with re-training, the physiological benefit remains to be determined.

An epigenetic memory at the CYP1A gene in cancer-resistant, pollution-adapted killifish.

Human exposure to polycyclic aromatic hydrocarbons (PAH) is a significant and growing public health problem. Frequent, high dose exposures are likely to increase due to a warming climate and increased frequency of large-scale wildfires. Here, we characterize an epigenetic memory at the cytochrome P450 1A (CYP1A) gene in a population of wild Fundulus heteroclitus that has adapted to chronic, extreme PAH pollution. In wild-type fish, CYP1A is highly induced by PAH. In PAH-tolerant fish, CYP1A induction is blunted. Since CYP1A metabolically activates PAH, this memory protects these fish from PAH-mediated cancer. However, PAH-tolerant fish reared in clean water recover CYP1A inducibility, indicating that blunted induction is a non-genetic memory of prior exposure. To explore this possibility, we bred depurated wild fish from PAH-sensitive and - tolerant populations, manually fertilized exposure-naïve embryos, and challenged them with PAH. We observed epigenetic control of the reversible memory of generational PAH stress in F1 PAH-tolerant embryos. Specifically, we observed a bivalent domain in the CYP1A promoter enhancer comprising both activating and repressive histone post-translational modifications. Activating modifications, relative to repressive ones, showed greater increases in response to PAH in sensitive embryos, relative to tolerant, consistent with greater gene activation. Also, PAH-tolerant adult fish showed persistent induction of CYP1A long after exposure cessation, which is consistent with defective CYP1A shutoff and recovery to baseline. Since CYP1A expression is inversely correlated with cancer risk, these results indicate that PAH-tolerant fish have epigenetic protection against PAH-induced cancer in early life that degrades in response to continuous gene activation.

Differential role of NMDA receptors in hippocampal-dependent spatial memory and plasticity in juvenile male and female rats.

Early life, or juvenility, stands out as the most pivotal phase in neurodevelopment due to its profound impact over the long-term cognition. During this period, significant changes are made in the brain's connections both within and between different areas, particularly in tandem with the development of more intricate behaviors. The hippocampus is among the brain regions that undergo significant postnatal remodeling, including dendritic arborization, synaptogenesis, the formation of complex spines and neuron proliferation. Given the crucial role of the hippocampus in spatial memory processing, it has been observed that spatial memory abilities continue to develop as the hippocampus matures, particularly before puberty. The N-methyl-d-aspartate (NMDA) type of glutamate receptor channel is crucial for the induction of activity-dependent synaptic plasticity and spatial memory formation in both rodents and humans. Although extensive evidence shows the role of NMDA receptors (NMDAr) in spatial memory and synaptic plasticity, the studies addressing the role of NMDAr in spatial memory of juveniles are sparse and mostly limited to adult males. In the present study, we, therefore, aimed to investigate the effects of systemic NMDAr blockade by the MK-801 on spatial memory (novel object location memory, OLM) and hippocampal plasticity in the form of long-term potentiation (LTP) of both male and female juvenile rats. Our results show the sex-dimorphic role of NMDAr in spatial memory and plasticity during juvenility, as systemic NMDAr blockade impairs the OLM and LTP in juvenile males without an effect on juvenile females. Taken together, our results demonstrate that spatial memory and hippocampal plasticity are NMDAr-dependent in juvenile males and NMDAr-independent in juvenile females. These sex-specific differences in the mechanisms of spatial memory and plasticity may imply gender-specific treatment for spatial memory disorders even in children.

PWAM-LSTM: prime word focused attention mechanism in long short term memory

PWAM-LSTM: prime word focused attention mechanism in long short term memory

Long-term saturated fat-enriched diets impair hippocampal learning and memory processes in a sex-dependent manner

Consumption of saturated fat-enriched diets during adolescence has been closely associated with the reduction of hippocampal synaptic plasticity and the impairment of cognitive function. Nevertheless, the effect of long-term intake of these foods has not yet been studied. In the present study, we have investigated the effect of a treatment, lasting for 40 weeks, with a diet enriched in saturated fat (SOLF) on i) spatial learning and memory, ii) hippocampal synaptic transmission and plasticity, and iii) hippocampal gene expression levels in aged male and female mice. Our findings reveal that SOLF has a detrimental impact on spatial memory and synaptic plasticity mechanisms, such as long-term potentiation (LTP), and downregulates Gria1 expression specifically in males. In females, SOLF downregulates the gene expression of Gria1/2/3 and Grin1/2A/2B glutamate receptor subunits as well as some proinflammatory interleukins. These findings highlight the importance of considering sex-specific factors when assessing the long-term effects of high-fat diets on cognition and brain plasticity.

Enhancing the Performance of Electrochemical RAM (ECRAM) through Modeling Guided Device Engineering

Traditionally, the only ions allowed or welcomed by the microelectronic industry are those that act as fixed dopants. The issue of mobile ions became welcome primarily due to the memristor technology however, thin film transistor technologies benefit from them too. The electrochemical (transistor) random access memory (ECRAM) is emerging as a promising building block for multi-level neuromorphic computing. Using FAB-compatible materials we construct the ECRAM using CuOx as the gate and ions source (Cu+). The morphology of the HfOx gate insulator layer is tuned to render it ion transporting such that it can act as a uniform electrolyte layer. Lastly, the channel material is WOx with tungsten metal as the source/drain contact.While there are several reports of ECRAM devices, the operation mechanisms are not fully known/understood thus withholding progress of this field. Using the Sentaurus device simulator by Synopsis, including the hydrogen diffusion module, we simulate the mixed ionic electronic operation of the device. We have recently reported that by fitting the simulation to the device performance, we could identify the potentiation mechanism (i.e., insulator charging) and the occurrence of copper plating that takes place under high Cu+ ion flux (as in fast charging of Li batteries).[1]In the first part of the talk, we will expand on the chemical-physics details of the ECRAM device mentioned above. Next, we will present a new device architecture where we remove the WOx layer and study the ionic-electronic conduction of a modified HfOx layer. By fine-tuning the stoichiometry of the HfOx layer, it assumes both roles of ionic electrolyte and electronic conducting channel. Following detailed modelling of the measured properties, we introduce AlO2 as an ionic barrier layer at the WOx/HfOx interface to enhance the ionic retention of the HfOx layer. Using the above three device architectures in conjunction with the mixed ionic-electronic device simulation we reveal the role of the two memory mechanisms: a) Electric field-activated ion transport and b) Structure-induced trapping by ion-barrier layers.Reference[1] Nir Tessler, Nayeon Kim, Heebum Kang, Jiyong Woo; Switching mechanisms of CMOS-compatible ECRAM transistors—Electrolyte charging and ion plating. J. Appl. Phys. 21 August 2023; 134 (7): 074501. https://doi.org/10.1063/5.0154153 Figure 1

Effects of Context Changes on Memory Reactivation.

While the influence of context on long-term memory (LTM) is well documented, its effects on the interaction between working memory (WM) and LTM remain less understood. In this study, we explored these interactions using a delayed match-to-sample task, where participants (6 males, 16 females) encountered the same target object across six consecutive trials, facilitating the transition from WM to LTM. During half of these target repetitions, the background color changed. We measured the WM storage of the target using the contralateral delay activity in electroencephalography. Our results reveal that task-irrelevant context changes trigger the reactivation of long-term memories in WM. This reactivation may be attributed to content-context binding in WM and hippocampal pattern separation.

A physical constraint-based machine learning model for shale oil production prediction

Shale oil has become a crucial unconventional resource, bolstering energy supply security, and it is important to accurately predict shale oil production dynamics. However, traditional prediction methods are faced with the challenges of large data fluctuations and numerous interference factors, which make accurate prediction difficult. This paper introduces a deep learning approach, specifically a physical constraint-based convolutional neural network combined with long short-term memory and attention mechanism (CNN-LSTM-AM), to predict shale oil production dynamics. Initially, the correlation analysis method is used to analyze the correlation strengths of features with the prediction target, and the data that are most strongly correlated with the daily oil production are screened out and preprocessed. Key features are then extracted; that is, the CNN layer's output is fed into the LSTM layer, the output of which informs a fully connected layer for time-series production predictions. The model integrates an AM to concentrate on vital data aspects. In the “five-linear flow” formula, the imbibition is hard-coded into the shale oil production prediction model. Physical constraints are introduced into the model. Data driven and empirical formulas are used to introduce a loss function as a constraint condition in the training process of a machine learning model to improve the interpretability and predictive ability of the model. During the tuning phase, Bayesian optimization is used to fine-tune the model's hyperparameters. Shapley additive explanation and accumulated local effects analyses are used to further assess and quantify the significance of the essential parameters, thus optimizing the prediction effectiveness.

Energy-Efficient Sleep Apnea Detection Using a Hyperdimensional Computing Framework Based on Wearable Bracelet Photoplethysmography.

Sleep apnea syndrome (SAS) is a common sleep disorder, which has been shown to be an important contributor to major neurocognitive and cardiovascular sequelae. Considering current diagnostic strategies are limited with bulky medical devices and high examination expenses, a large number of cases go undiagnosed. To enable large-scale screening for SAS, wearable photoplethysmography (PPG) technologies have been used as an early detection tool. However, existing algorithms are energy-intensive and require large amounts of memory resources, which are believed to be the major drawbacks for further promotion of wearable devices for SAS detection. In this paper, an energy-efficient method of SAS detection based on hyperdimensional computing (HDC) is proposed. Inspired by the phenomenon of chunking in cognitive psychology as a memory mechanism for improving working memory efficiency, we proposed a one-dimensional block local binary pattern (1D-BlockLBP) encoding scheme combined with HDC to preserve dominant dynamical and temporal characteristics of pulse rate signals from wearable PPG devices. Our method achieved 70.17 % accuracy in sleep apnea segment detection, which is comparable with traditional machine learning methods. Additionally, our method achieves up to 67× lower memory footprint, 68× latency reduction, and 93× energy saving on the ARM Cortex-M4 processor. The simplicity of hypervector operations in HDC and the novel 1D-BlockLBP encoding effectively preserve pulse rate signal characteristics with high computational efficiency. This work provides a scalable solution for long-term home-based monitoring of sleep apnea, enhancing the feasibility of consistent patient care.

A Spatial-Temporal Video Quality Assessment Method via Comprehensive HVS Simulation.

The quality of videos is the primary concern of video service providers. Built upon deep neural networks, video quality assessment (VQA) has rapidly progressed. Although existing works have introduced the knowledge of the human visual system (HVS) into VQA, there are still some limitations that hinder the full exploitation of HVS, including incomplete modeling with few HVS characteristics and insufficient connection among these characteristics. In this article, we present a novel spatial-temporal VQA method termed HVS-5M, wherein we design five modules to simulate five characteristics of HVS and create a bioinspired connection among these modules in a cooperative manner. Specifically, on the side of the spatial domain, the visual saliency module first extracts a saliency map. Then, the content-dependency and the edge masking modules extract the content and edge features, respectively, which are both weighted by the saliency map to highlight those regions that human beings may be interested in. On the other side of the temporal domain, the motion perception module extracts the dynamic temporal features. Besides, the temporal hysteresis module simulates the memory mechanism of human beings and comprehensively evaluates the video quality according to the fusion features from the spatial and temporal domains. Extensive experiments show that our HVS-5M outperforms the state-of-the-art VQA methods. Ablation studies are further conducted to verify the effectiveness of each module toward the proposed method. The source code is available at https://github.com/GZHU-DVL/HVS-5M.

The avidity of virus-specific antibodies obtained from <i>in vitro</i> stimulated memory b cells does not change one month after booster with Sputnik V or Comirnaty

The protective properties of long-term immunological memory after vaccination against COVID-19 are characterized by the neutralizing activity of serum antibodies and antibodies secreted by memory B cells upon repeated encounter with the antigen. Somatic hypermutations occurring in the immunoglobulin genes of memory B cells are one of the mechanisms for increasing the affinity of antibodies. At the moment, the effect of booster vaccination against COVID-19 with vector vaccines, on the maturation of memory B cells remains poorly understood. The purpose of this work was to determine how COVID-19 booster affects the affinity of RBD-specific IgG antibodies secreted by memory B cells. B lymphocytes were isolated from peripheral mononuclear blood cells of volunteers who had been revaccinated against COVID-19 with Sputnik V or Comirnaty. B cells were stimulated in vitro with CD40L expressed on the surface of A549 feeder cells and IL-21. Supernatants were concentrated 8-fold using centrifugal concentrators. In the obtained supernatants from stimulated memory B cells, the level of IgG antibodies specific to wild-type RBD was determined by enzyme-linked immunosorbent assay (ELISA). To determine the avidity index, ELISA with 7M urea was provided. It was shown that despite a general increase in the amount of antigen-specific IgG antibodies obtained from stimulated memory B cells, there was no change in the avidity of these antibodies one month after booster in both groups of donors. The obtained results contribute to the understanding of the mechanisms of memory B cell maturation after booster vaccinations against COVID-19 and may be useful for deciding on the strategy of booster vaccination.

Oxytocin induces the formation of distinctive cortical representations and cognitions biased toward familiar mice

Social recognition is essential for the formation of social structures. Many times, recognition comes with lesser exploration of familiar animals. This lesser exploration has led to the assumption that recognition may be a habituation memory. The underlying memory mechanisms and the thereby acquired cortical representations of familiar mice have remained largely unknown, however. Here, we introduce an approach directly examining the recognition process from volatile body odors among male mice. We show that volatile body odors emitted by mice are sufficient to identify individuals and that more salience is assigned to familiar mice. Familiarity is encoded by reinforced population responses in two olfactory cortex hubs and communicated to other brain regions. The underlying oxytocin-induced plasticity promotes the separation of the cortical representations of familiar from other mice. In summary, neuronal encoding of familiar animals is distinct and utilizes the cortical representational space more broadly, promoting storage of complex social relationships.

Mild neonatal hypoxia disrupts adult hippocampal learning and memory and is associated with CK2-mediated dysregulation of synaptic calcium-activated potassium channel KCNN2.

Although nearly half of preterm survivors display persistent neurobehavioral dysfunction including memory impairment without overt gray matter injury, the underlying mechanisms of neuronal or glial dysfunction, and their relationship to commonly observed cerebral white matter injury are unclear. We developed a mouse model to test the hypothesis that mild hypoxia during preterm equivalence is sufficient to persistently disrupt hippocampal neuronal maturation related to adult cellular mechanisms of learning and memory. Methods: Neonatal (P2) mice were exposed to mild hypoxia (8%O 2 ) for 30 min and evaluated for acute injury responses or survived until adulthood for assessment of learning and memory and hippocampal neurodevelopment. Neonatal mild hypoxia resulted in clinically relevant oxygen desaturation and tachycardia without bradycardia and was not accompanied by cerebral gray or white matter injury. Neonatal hypoxia exposure was sufficient to cause hippocampal learning and memory deficits and abnormal maturation of CA1 neurons that persisted into adulthood. This was accompanied by reduced hippocampal CA3-CA1 synaptic strength and LTP and reduced synaptic activity of calcium-sensitive SK2 channels, key regulators of spike timing dependent neuroplasticity, including LTP. Structural illumination microscopy revealed reduced synaptic density, but intact SK2 localization at the synapse. Persistent loss of SK2 activity was mediated by altered casein kinase 2 (CK2) signaling. Clinically relevant mild hypoxic exposure in the neonatal mouse is sufficient to produce morphometric and functional disturbances in hippocampal neuronal maturation independently of white matter injury. Additionally, we describe a novel persistent mechanism of potassium channel dysregulation after neonatal hypoxia. Collectively our findings suggest an unexplored explanation for the broad spectrum of neurobehavioral, cognitive and learning disabilities that paradoxically persist into adulthood without overt gray matter injury after preterm birth.

Volleyball training video classification description using the BiLSTM fusion attention mechanism

This study aims to explore methods for classifying and describing volleyball training videos using deep learning techniques. By developing an innovative model that integrates Bi-directional Long Short-Term Memory (BiLSTM) and attention mechanisms, referred to BiLSTM-Multimodal Attention Fusion Temporal Classification (BiLSTM-MAFTC), the study enhances the accuracy and efficiency of volleyball video content analysis. Initially, the model encodes features from various modalities into feature vectors, capturing different types of information such as positional and modal data. The BiLSTM network is then used to model multi-modal temporal information, while spatial and channel attention mechanisms are incorporated to form a dual-attention module. This module establishes correlations between different modality features, extracting valuable information from each modality and uncovering complementary information across modalities. Extensive experiments validate the method's effectiveness and state-of-the-art performance. Compared to conventional recurrent neural network algorithms, the model achieves recognition accuracies exceeding 95 % under Top-1 and Top-5 metrics for action recognition, with a recognition speed of 0.04 s per video. The study demonstrates that the model can effectively process and analyze multimodal temporal information, including athlete movements, positional relationships on the court, and ball trajectories. Consequently, precise classification and description of volleyball training videos are achieved. This advancement significantly enhances the efficiency of coaches and athletes in volleyball training and provides valuable insights for broader sports video analysis research.

F-PICNN: A physics-informed convolutional neural network for partial differential equations with space-time domain

The physics and interdisciplinary problems in science and engineering are mainly described as partial differential equations (PDEs). Recently, a novel method using physics-informed neural networks (PINNs) to solve PDEs by employing deep neural networks with physical constraints as data-driven models has been pioneered for surrogate modelling and inverse problems. However, the original PINNs based on fully connected neural networks pose intrinsic limitations and poor performance for the PDEs with nonlinearity, drastic gradients, multiscale characteristics or high dimensionality in which the complex features are hard to capture. This leads to difficulties in convergence to correct solutions and high computational costs. To address the above problems, in this paper, a novel physics-informed convolutional neural network framework based on finite discretization schemes with a stack of a series of nonlinear convolutional units (NCUs) for solving PDEs in the space-time domain without any labelled data (f-PICNN) is proposed, in which the memory mechanism can considerably speed up the convergence. Specifically, the initial conditions (ICs) are hard-encoded into the network as the first time-step solution and used to extrapolate the next time-step solution. The Dirichlet boundary conditions (BCs) are constrained by soft BC enforcement while the Neumann BCs are hard enforced. Furthermore, the loss function is designed as a set of discretized PDE residuals and optimized to conform to physics laws. Finally, the proposed auto-regressive model has been proven to be effective in a wide range of 1D and 2D nonlinear PDEs in both space and time under different finite discretization schemes (e.g., Euler, Crank Nicolson and fourth-order Runge-Kutta). The numerical results demonstrate that the proposed framework not only shows the ability to learn the PDEs efficiently but also provides an opportunity for greater conceptual simplicity, and potential for extrapolation from learning the PDEs using a limited dataset.

Comparative proteomics in tall fescue to reveal underlying mechanisms for improving Photosystem II thermotolerance during heat stress memory

BackgroundThe escalating impacts of global warming intensify the detrimental effects of heat stress on crop growth and yield. Among the earliest and most vulnerable sites of damage is Photosystem II (PSII). Plants exposed to recurring high temperatures develop heat stress memory, a phenomenon that enables them to retain information from previous stress events to better cope with subsequent one. Understanding the components and regulatory networks associated with heat stress memory is crucial for the development of heat-resistant crops.ResultsPhysiological assays revealed that heat priming (HP) enabled tall fescue to possess higher Photosystem II photochemical activity when subjected to trigger stress. To investigate the underlying mechanisms of heat stress memory, we performed comparative proteomic analyses on tall fescue leaves at S0 (control), R4 (primed), and S5 (triggering), using an integrated approach of Tandem Mass Tag (TMT) labeling and Liquid Chromatography-Mass Spectrometry. A total of 3,851 proteins were detected, with quantitative information available for 3,835 proteins. Among these, we identified 1,423 differentially abundant proteins (DAPs), including 526 proteins that were classified as Heat Stress Memory Proteins (HSMPs). GO and KEGG enrichment analyses revealed that the HSMPs were primarily associated with the “autophagy” in R4 and with “PSII repair”, “HSP binding”, and “peptidase activity” in S5. Notably, we identified 7 chloroplast-localized HSMPs (HSP21, DJC77, EGY3, LHCA4, LQY1, PSBR and DEGP8, R4/S0 > 1.2, S5/S0 > 1.2), which were considered to be effectors linked to PSII heat stress memory, predominantly in cluster 4. Protein-protein interaction (PPI) analysis indicated that the ubiquitin-proteasome system, with key nodes at UPL3, RAD23b, and UCH3, might play a role in the selective retention of memory effectors in the R4 stage. Furthermore, we conducted RT-qPCR validation on 12 genes, and the results showed that in comparison to the S5 stage, the R4 stage exhibited reduced consistency between transcript and protein levels, providing additional evidence for post-transcriptional regulation in R4.ConclusionsThese findings provide valuable insights into the establishment of heat stress memory under recurring high-temperature episodes and offer a conceptual framework for breeding thermotolerant crops with improved PSII functionality.

Graph recurrent neural networks-integrated real-time prediction of key displacements for fire-induced collapse early warning of steel frames

This study proposes the FRAME-Net, which can extract spatial-temporal features of discrete structures like planar steel frames, for early warning systems of fire-induced building collapse. FRAME-Net, based on graph neural networks and recurrent neural networks, seamlessly integrates feature extraction techniques, advanced graph convolution operations, and an enhanced graph long short-term memory mechanism, resolving the critical issue that the actual state of burning buildings, e.g., fire scenario, load case, material properties, etc., cannot be identified accurately and rapidly. The temperatures of steel components and easy-to-measure displacements are inputs, and hard-to-measure displacements at the top and interior of steel frames under fire can be predicted in real time. Numerical examples indicate that the trained agent accurately predicts the hard-to-measure displacements for the trained steel frame under completely unknown datasets, showcasing its potential in real-world unknown fire scenarios. Besides, the trained intelligent agent can be directly applied to untrained steel frames with different topological forms without re-training, which significantly reduces computational costs and is beneficial for swift emergency responses. Subsequently, satisfactory results predicted by the trained agent without re-training for the steel frame under the real fire test further underscores the promising generalization capability of the trained agent to adapt to unknown fire scenarios in real-world steel frames. Finally, the feasibility of the proposed framework to facilitate the implementation of the early-warning systems of fire-induced collapse is proved by comparing the predicted and actual remaining building collapse time, thereby assisting in reducing casualties during the rescue process and contributing to smart firefighting advancements.

OptiLoc: Integrating SCSO and DV-Hop for wireless sensor network localization with application to disease forecasting in cattle farm monitoring

Disease prediction and management in cattle farming present significant challenges due to the complex interplay of environmental factors, disease transmission dynamics and the need for accurate localization of disease-carrying cattle. Traditional approaches often fall short in optimizing anchor node selection and accurately forecasting disease occurrences. To address these limitations, we propose OptiLoc, a novel framework that integrates advanced optimization techniques and disease forecasting algorithms. The proposed OptiLoc framework leverages Enhanced Sand Cat Swarm Optimization (SCSO) for optimizing anchor node selection within the farm area network. By dynamically adjusting the exploration rate and incorporating an opposition-based memory mechanism, OptiLoc ensures thorough exploration of the search space while preventing premature convergence. Furthermore, the multi-objective optimization aspect of SCSO considers factors such as coverage, energy efficiency, and connectivity, resulting in optimal anchor node placement. Additionally, OptiLoc integrates disease forecasting capabilities using logistic regression classifiers with dictionary learning. By extracting meaningful features from sensor data and training linear classifiers, OptiLoc predicts disease outbreaks among the cattle population. These predictions are seamlessly integrated into the optimization process through an expanded fitness function, which considers disease-related parameters alongside traditional optimization objectives. Furthermore, OptiLoc enhances disease management and control efforts by localizing sensor nodes using the DV-Hop method. By estimating sensor node positions based on hop distances to anchor nodes and dynamically calibrating hop distances based on real-time data on disease prevalence and environmental factors, OptiLoc achieves precise localization results. The proposed OptiLoc framework offers several advantages over traditional approaches. It optimizes anchor node placement, accurately forecasts disease outbreaks, and enhances localization accuracy, thereby empowering farmers with actionable insights to safeguard the health and well-being of their livestock. Performance evaluation reveals for the anchor nodes, it measures a mean localization error of 3.2% to 9.1% with a mean error of 5.91%. Similarly, for the sensor nodes, the mean localization error is 3.45% to 8.25% with a mean error of 5.445%. The AUC of the logistic regression classifier with dictionary learning is 0.97. Through its comprehensive approach, OptiLoc represents a significant advancement in optimizing WSNs in agricultural environments.

Artificial Neural Network Learning, Attention, and Memory

The learning equations of an ANN are presented, giving an extremely concise derivation based on the principle of backpropagation through the descendent gradient. Then, a dual network is outlined acting between synapses of a basic ANN, which controls the learning process and coordinates the subnetworks selected by attention mechanisms toward purposeful behaviors. Mechanisms of memory and their affinity with comprehension are considered, by emphasizing the common role of abstraction and the interplay between assimilation and accommodation, in the spirit of Piaget’s analysis of psychological acquisition and genetic epistemology. Learning, comprehension, and knowledge are expressed as different levels of organization of informational processes inside cognitive systems. It is argued that formal analyses of cognitive artificial systems could shed new light on typical mechanisms of “natural intelligence” and, in a specular way, that models of natural cognition processes could promote further developments of ANN models. Finally, new possibilities of chatbot interaction are briefly discussed.