Mnemonic Research Articles

Recurrent Neural Networks for Image Captioning: A Case Study with LSTM

This research investigates the viability of Long Short-Term Memory (LSTM) systems, a subtype of Recurrent Neural Networks (RNNs), for picture captioning. Leveraging the MS COCO dataset, the study compares the execution of LSTM-based RNNs with Vanilla RNN, Gated Recurrent Unit (GRU), consideration components, and transformer-based models. Experimental comes about to illustrate that the LSTM-based RNN shows competitive execution, accomplishing a BLEU-4 score of 0.72, a METEOR score of 0.68, and a CIDEr score of 2.1. The comparative investigation uncovers its prevalence over Vanilla RNN and GRU, highlighting its capability to capture long-range conditions inside successive picture information. Moreover, the study investigates the effect of consideration instruments and transformer designs, exhibiting their potential improvements in the context-aware caption era. The transformer-based show outflanks all other models, accomplishing a BLEU-4 score of 0.78, a METEOR score of 0.72, and a CIDEr score of 2.5. The findings give important bits of knowledge toward the creating scene of picture captioning strategy, which makes LSTM-based RNNs solid and productive approaches for capturing worldly groupings in visual substance. In achieving these, the study provides a framework for future developments in hybrid models and manufacturing processes that push boundaries of smart image perception and understanding

Heterogenous effect of early adulthood stress on cognitive aging and synaptic function in the dentate gyrus.

Cognitive aging widely varies among individuals due to different stress experiences throughout the lifespan and vulnerability of neurocognitive mechanisms. To understand the heterogeneity of cognitive aging, we investigated the effect of early adulthood stress (EAS) on three different hippocampus-dependent memory tasks: the novel object recognition test (assessing recognition memory: RM), the paired association test (assessing episodic-like memory: EM), and trace fear conditioning (assessing trace memory: TM). Two-month-old rats were exposed to chronic mild stress for 6 weeks and underwent behavioral testing either 2 weeks or 20 months later. The results show that stress and aging impaired different types of memory tasks to varying degrees. RM is affected by combined effect of stress and aging. EM became less precise in EAS animals. TM, especially the contextual memory, showed impairment in aging although EAS attenuated the aging effect, perhaps due to its engagement in emotional memory systems. To further explore the neural underpinnings of these multi-faceted effects, we measured long-term potentiation (LTP), neural density, and synaptic density in the dentate gyrus (DG). Both stress and aging reduced LTP. Additionally, the synaptic density per neuron showed a further reduction in the stress aged group. In summary, EAS modulates different forms of memory functions perhaps due to their substantial or partial dependence on the functional integrity of the hippocampus. The current results suggest that lasting alterations in hippocampal circuits following EAS could potentially generate remote effects on individual variability in cognitive aging, as demonstrated by performance in multiple types of memory.

A thematic analysis of the introduction of smart-hub technology to a rural Psychiatry of Old Age Service during Covid-19 lockdowns

IntroductionThe use of smart technology in supporting older adults is a growing field of research. However, there is little qualitative research on the experiences of patients utilizing this technology, particularly those attending psychiatry services.ObjectivesTo explore the experiences of staff and patients utilizing smart-hubs implemented during the Covid-19 pandemic to provide remote audio/visual communication and smart AI personal assistant technology for the management of patients in a rural Psychiatry of Old Age service.MethodsSmart hubs were installed in patient homes and in the Psychiatry of Old Age base during the Covid-19 pandemic when lockdown restrictions limited in-person service provision. Patients and staff utilized the smart hubs for its assistive technology and to engage with each other. Semi-structured qualitative interviews were conducted of 10 staff and 15 patients at 6-12 months following the introduction of the smart hubs and thematic analysis was conducted to generate themes.ResultsThree themes were generated from the thematic analysis: 1) Openness to SMART hub technology, 2) Getting set-up and 3) Keeping SMART. The majority of participants did not have much experience using smart technology prior to the intervention. However, many participants reported that they would be comfortable using technology. The Covid-19 pandemic contributed to the rapid adoption of this intervention within the service with mixed views regarding the smart hub prior to implementation. The role of key individuals such as staff and family was highlighted in supporting older persons with setting-up the smart hub. Technical needs included the need for a strong internet connection and technical limitations were driven by privacy, cost and regional considerations. Many patients were able to utilize the smart hub independently to access interests, therapeutic activities and as a memory aid. The smart hub offered a novel way to connect to services and families and was also seen as a companion by some patients and staff to help address loneliness and isolation. The majority of participants found the use of smart hubs acceptable and were willing to utilize the smart hub in the future as an adjunct to face to face psychiatric interventions. However, suggestions for future use included the need for additional training as users felt that there was more they could do with the smart hub, continued support to manage any challenges and improved information leaflets to better engage users.ConclusionsSmart hub technology offers an alternate means of providing remote and inclusive psychiatric care to older patients unable to access services in person and at risk of deterioration without intervention in the community.Disclosure of InterestNone Declared

Stimulation of the collective memory of the 1999 Turkey earthquake through the Turkish media coverage of the 2023 earthquake

Turkey has been struck by several powerful earthquakes. Since the 1999 earthquake was the most recent and devastating earthquake before the last one happened in February 2023, many of these media channels’ depictions of the 2023 earthquake in Turkey may have been impacted by the collective memory of the 1999 earthquake. Collective memory of disasters and conflicts frequently takes on special significance as a mechanism for society to cope with the catastrophic events they have witnessed. Collective memory aids societies in dealing with the consequences of such events by giving a feeling of continuity as well as a structure for interpreting and comprehending what occurred. The media and social media are important in developing and conveying collective memory. They play an important role in framing events, transmitting details, and providing a forum for public debate. Social media, in addition to traditional media, has emerged as an innovative platform for the construction and diffusion of collective memory. The purpose of this study is to investigate whether the Turkish media depicted the collective memory of the 1999 earthquake in the aftermath of the 2023 earthquake. If the collective memory of the 1999 earthquake is invoked in media coverage of the 2023 earthquake, how is it depicted in terms of lessons learned, public response, and influence on Turkish society? The study’s findings indicate that the analysis of Turkish media coverage pertaining to the 2023 earthquake has demonstrated a restricted collective recollection of the 1999 earthquake.

System and Design Technology Co-Optimization of SOT-MRAM for High-Performance AI Accelerator Memory System

System and Design Technology Co-Optimization of SOT-MRAM for High-Performance AI Accelerator Memory System

Design and Implementation of Low Power and High Data Rate Edge Coded Signaling Architecture for IoT Devices

Abstract Edge coded signaling (ECS) is a recently introduced communication protocol designed for single-channel signaling among constrained internet of things (IoT) nodes. This study aims to enhance the data rate of ECS, without increasing the clock frequency or power consumption. The goal is to develop an improved protocol that maintains the inherent advantages of ECS while achieving a higher data rate. The proposed solution, double data rate ECS (DDR-ECS), leverages both pulse edges of the ECS pulse stream for information encoding, drawing an analogy to DDR memory systems. The study presents a detailed design of low-power architecture for the DDR-ECS transceiver. This design was synthesized using a 65 nm ASIC process to evaluate its performance in terms of power consumption, gate count, and data rate. The synthesis results demonstrate that the DDR-ECS transceiver preserves the key features of ECS, including tolerance to clock frequency variations and compatibility with lightweight cryptographic algorithms. It effectively doubles the dynamic data rate to the range of 12–73.5 Mb/s while consuming an equivalent power of 13 μW and occupying a compact form factor of only 1,755 gates. The introduction of DDR-ECS represents a significant advancement in the ECS protocol, offering a novel approach to doubling the data rate without increasing the clock frequency or power envelope. This innovation retains the simplicity and efficiency of ECS, providing an enhanced communication solution for constrained IoT nodes.

The Effects of Prospective Memory Intervention on Language Ability and Activities of Daily Living in the Elderly with Alzheimer’s Disease

Objectives: Alzheimer’s Disease (AD) is characterized by an early decline in memory and cognitive functions, leading to impairments in language abilities and difficulties in communication as the disease progresses. Therefore, the need for early diagnosis and intervention in AD is considered critical; and maintaining and improving Prospective Memory (PM) is reported to aid in enhancing complex cognitive functions. This study aimed to explore the changes observed before and after PM intervention using External Memory Aids (EMA) in elderly individuals with AD.Methods: This study was conducted with 7 elderly individuals with mild AD, over three weeks with 12 sessions, four times a week. The intervention was based on a protocol consisting of three stages using the Prospective Memory Task (PMT) developed for this research: 1) reading texts, 2) answering wh-questions, and 3) utilizing EMA.Results: Significant improvements were showed after the intervention in performance of intervention and non-intervention PMT, immediate recall of words and story, naming, and K-MMSE, indicating the generalized effects of the intervention. However, there were no significant improvements in delayed recall, verb fluency and Activities of Daily Living.Conclusion: The use of EMA in PM intervention appeared to positively impact the language abilities and overall cognitive functions of elderly AD patients, suggesting that the activated PM through the intervention could lead to improvements in general cognitive functions.

Angara Style Rock Art: The Evolution of a Regional Emblematic and Syncretic Style

Rebutting previous claims, the paper employs comparative stylistic analysis and palaeoenvironmental data to argue that Angara style rock art originated in the Mongolian Altai during the Upper Palaeolithic (13,000–10,300 bp) where it evolved in situ. Around 8200–7300 bp, drought forced the hunter-gatherers who created Angara style rock art to migrate to the Upper Yenisey and the Selenga and Angara basins. When drought impacted that area c. 7500–7000 bp, Kotoi (Ket) culture descendants sought refuge in the resource-rich Minusinsk Basin. On the Middle Yenisey River, Angara style rock art served as a mnemonic device that encoded the syncretism of proto Ket and Evenki cosmologies and beliefs resulting from their social alliance.

O versiune românească necunoscută a Imnului Acatist din a doua jumătate a secolului al XVII-lea

The present paper aims to describe, date and localize the Akathistos Hymn found in an unknown Romanian manuscript preserved at Muzeul Olteniei (Craiova): Rom MS I 529. The codex contains the Psalter (accompanied by the Odes), the Akathistos Hymn, the Paraklesis to Theotokos and the Holy Apostle. These texts were copied by father Theofil, the scribe of the manuscript. These works are important due to two outstanding reasons: they contain Slavic passages and multiple translation variants inserted inside the text using numbers or other graphic devices. Both of these features can be found in the manuscripts produced in Bisericani Monastery (Moldova) in the first part of the 17 th century. As a consequence, the codex transcribed by Teofil was analysed in the context of the production of the scriptorium from Bisericani. Teofil must have been part of the Moldavian monastic community and his manuscript must have been copied there. Textual, contextual, linguistic, philological and palaeographical evidence are given in this sense. The study of the watermarks (Torch with letters HSE, unicorn encircled by letters, Allmodpappier, horn and the ligature 4AD), their types and subtypes, brought to light the fact that the Romanian manuscript was copied between 1680-1690. The philological comparison between the Biblical and Marian texts from these manuscripts and the oldest ones from Bisericani (Rom MS BAR 540 and 170) stressed the link between the former and the latter. On top of that, the comparison offered new insight into how the multiple translation variants were more or less preserved, hinting to an early contamination between different manuscripts. 3 The Slavic passages (mostly headings and initial lines) preserved in the Romanian texts were studied as a mnemonic device used by the monks from Bisericani Monastery who, according to the liturgic program of the community, were used to chanting the Akathistos Hymn in Slavonic.

I remember it now, so I'll remember it later: Working memory strength guides predictions for long-term memory performance.

Judgments of learning (JOLs) are assumed to be made inferentially, based on cues. This cue-utilization approach substituted the theory that memory strength guides JOLs. The rejection of this theory ignores the existence of two memory systems: working memory (WM), which holds representations immediately accessible, and long-term memory (LTM), which is a permanent store. By manipulating and measuring WM strength, we tested a revised version of the memory-strength theory in which JOLs are guided by WM representations. In Experiment 1, participants memorized sequences of two or four colored objects, then they provided JOLs for an LTM test of these objects, and performed a WM test on the objects' colors. After learning 200 objects, the LTM test followed. Sequence-length affected WM, but not LTM performance. JOLs, however, were higher for sequences of two than for four objects and correlated higher with WM than LTM performance. We replicated these results with a simultaneous presentation of the objects (Experiment 2), in the absence of a WM test (Experiment 3), and in a word-pair task (Experiment 4). Overall, our findings are consistent with the revised memory-strength theory. WM strength should therefore be considered when examining the factors guiding JOLs.

Irreversible atrophy in memory brain regions over 7 years is predicted by glycemic control in type 2 diabetes without mild cognitive impairment.

Memory-related impairments in type 2 diabetes may be mediated by insulin resistance and hyperglycemia. Previous cross-sectional studies have controversially suggested a relationship between metabolic control and a decrease in hippocampal volumes, but only longitudinal studies can test this hypothesis directly. We performed a longitudinal morphometric study to provide a direct test of a possible role of higher levels of glycated hemoglobin with long term brain structural integrity in key regions of the memory system - hippocampus, parahippocampal gyrus and fusiform gyrus. Grey matter volume was measured at two different times - baseline and after ~7 years. We found an association between higher initial levels of HbA1C and grey matter volume loss in all three core memory regions, even in the absence of mild cognitive impairment. Importantly, these neural effects persisted in spite of the fact that patients had significantly improved their glycemic control. This suggests that early high levels of HbA1c might be irreversibly associated with subsequent long-term atrophy in the medial temporal cortex and that early intensive management is critical.

Dissociable Contributions of the Medial Parietal Cortex to Recognition Memory.

Human neuroimaging studies of episodic memory retrieval routinely observe the engagement of specific cortical regions beyond the medial temporal lobe. Of these, medial parietal cortex (MPC) is of particular interest given its distinct functional characteristics during different retrieval tasks. Specifically, while recognition and autobiographical recall tasks are both used to probe episodic retrieval, these paradigms consistently drive distinct spatial patterns of response within MPC. However, other studies have emphasized alternate MPC functional dissociations in terms of brain network connectivity profiles or stimulus category selectivity. As the unique contributions of MPC to episodic memory remain unclear, adjudicating between these different accounts can provide better consensus regarding MPC function. Therefore, we used a precision-neuroimaging dataset (7T functional magnetic resonance imaging) to examine how MPC regions are differentially engaged during recognition memory and how these task-related dissociations may also reflect distinct connectivity and stimulus category functional profiles. We observed interleaved, though spatially distinct, subregions of MPC where responses were sensitive to either recognition decisions or the semantic representation of stimuli. In addition, this dissociation was further accentuated by functional subregions displaying distinct profiles of connectivity with the hippocampus during task and rest. Finally, we show that recent observations of dissociable person and place selectivity within the MPC reflect category-specific responses from within identified semantic regions that are sensitive to mnemonic demands. Together, by examining precision functional mapping within individuals, these data suggest that previously distinct observations of functional dissociation within MPC conform to a common principle of organization throughout hippocampal-neocortical memory systems.

Evolving Parameterized Prompt Memory for Continual Learning

Recent studies have demonstrated the potency of leveraging prompts in Transformers for continual learning (CL). Nevertheless, employing a discrete key-prompt bottleneck can lead to selection mismatches and inappropriate prompt associations during testing. Furthermore, this approach hinders adaptive prompting due to the lack of shareability among nearly identical instances at more granular level. To address these challenges, we introduce the Evolving Parameterized Prompt Memory (EvoPrompt), a novel method involving adaptive and continuous prompting attached to pre-trained Vision Transformer (ViT), conditioned on specific instance. We formulate a continuous prompt function as a neural bottleneck and encode the collection of prompts on network weights. We establish a paired prompt memory system consisting of a stable reference and a flexible working prompt memory. Inspired by linear mode connectivity, we progressively fuse the working prompt memory and reference prompt memory during inter-task periods, resulting in continually evolved prompt memory. This fusion involves aligning functionally equivalent prompts using optimal transport and aggregating them in parameter space with an adjustable bias based on prompt node attribution. Additionally, to enhance backward compatibility, we propose compositional classifier initialization, which leverages prior prototypes from pre-trained models to guide the initialization of new classifiers in a subspace-aware manner. Comprehensive experiments validate that our approach achieves state-of-the-art performance in both class and domain incremental learning scenarios.

Mnemonics – Known and Unknown

The given article deals with the issue of mnemonics, their types and their application in the learning process and daily activities. Examples in several languages and for different study subjects and areas are covered. Mnemonic techniques proposed by artificial intelligence are reviewed. The work presents the results of a survey among 301 participants aimed at revealing people‘s awareness of the possibilities of mnemonics and their propensity to use them. The idea that mnemonics in school should be considered from two sides – as a finished product on the one hand and as an author‘s realization on the other hand – has been revealed. Creating your own mnemonics involves high cognitive levels and important soft skills.

A content analysis of medication adherence material in patient educational resources about gout.

This study aimed to investigate how medication adherence is addressed in online gout resources in six countries. We investigated how often adherence was referred to, the strategies suggested to improve patient adherence, and the types of nonadherence that were targeted. We also examined the readability of the adherence material. A content analysis was conducted on 151 online gout resources from medical and health organisations in six predominantly English-speaking countries. Two reviewers coded the content of the websites into categories (kappa 0.80). The analysis involved coding the resources for reasons for nonadherence, and adherence-promoting strategies. Flesch-Kincaid Reading Ease scores and word count were also computed. Out of 151 websites examined, 77 websites discussed medication adherence (51%), with intentional nonadherence being more prevalent than unintentional nonadherence. 67 websites targeted different types of nonadherence, including drug-specific concerns (50%), misconceptions of gout curability and the necessity of medication (16%), forgetfulness (16%), and other practical challenges (5%). Strategies to promote adherence were found in one-third of the websites, with medication education being the most prevalent strategy (17%), followed by healthcare provider engagement (13%) and memory aid strategies (6%). On average, about 11% of the words (89.27, SD = 76.35) in the entire document were focused on adherence. Difficult reading comprehension was found in one-fifth of adherence-related websites. Findings reveal limited medication adherence coverage and narrow strategies in online gout resources. Improved adherence portrayal is needed for effective gout management through comprehensive strategies and clear, understandable information.

Detection and confirmation of electricity thefts in Advanced Metering Infrastructure by Long Short-Term Memory and fuzzy inference system models

The successful implementation of Smart Grids heavily relies on energy efficiency, particularly through the Advanced Metering Infrastructure (AMI) and Smart Electricity Meters (SEM). However, cyber-attacks pose a threat to SEM, with electricity theft being a primary motivation. Despite the valuable data provided by SEM for analytical purposes, existing methods to identify theft involve cumbersome and costly on-site inspections. This research proposes an electricity theft detection model using the Long Short-Term Memory (LSTM) network. The model employs a collective anomaly approach, defining prediction errors through a threshold and forecast horizon. Suspicious consumption profiles are analysed, and a fuzzy inference system (FIS) implemented in MATLAB 2021b is used to model security risks based on these profiles. The study utilizes energy consumption data from four diverse consumer profiles (consumers 1, 2, 3, and 4) to develop consumer-specific LSTM models for detection and an FIS model for confirmation. Tampered consumer data is identified and confirmed based on selected AMI parameters. While all consumers exhibit suspicious profiles at times, only consumers 2 and 3 are confirmed as engaging in electricity theft. This research provides a robust approach to detecting and verifying fraudulent consumption profiles within the context of AMI, offering a more reliable dimension to theft detection and confirmation.

A new thread-level speculative automatic parallelization model and library based on duplicate code execution

Loop-efficient automatic parallelization has become increasingly relevant due to the growing number of cores in current processors and the programming effort needed to parallelize codes in these systems efficiently. However, automatic tools fail to extract all the available parallelism in irregular loops with indirections, race conditions or potential data dependency violations, among many other possible causes. One of the successful ways to automatically parallelize these loops is the use of speculative parallelization techniques. This paper presents a new model and the corresponding C++ library that supports the speculative automatic parallelization of loops in shared memory systems, seeking competitive performance and scalability while keeping user effort to a minimum. The primary speculative strategy consists of redundantly executing chunks of loop iterations in a duplicate fashion. Namely, each chunk is executed speculatively in parallel to obtain results as soon as possible and sequentially in a different thread to validate the speculative results. The implementation uses C++11 threads and it makes intensive use of templates and advanced multithreading techniques. An evaluation based on various benchmarks confirms that our proposal provides a competitive level of performance and scalability.

The Impact of Digital Technologies on Memory and Memory Studies

With the widespread integration of smartphones, computers, and the internet, information access and processing have undergone significant changes. This paper investigates both positive and negative implications, acknowledging the extension of cognitive capacities through easy access to vast databases and external memory aids while also addressing concerns about diminished memory consolidation and reliance on shallow encoding strategies. Examining the interdisciplinary field of memory studies, the study also highlights collaborative efforts among scholars in psychology, neuroscience, sociology, and information science to comprehend the impact of digital technologies on memory, and emphasizes the challenges and future directions in memory research, including issues like digital amnesia, information overload, and privacy concerns. Overall, the paper underscores the need for understanding the relationship between human memory and digital tools, enabling the development of strategies to enhance memory, counteract potential adverse effects, and promote a balanced utilization of digital resources in memory-related tasks.

Studying the Robustness of Anti-Adversarial Federated Learning Models Detecting Cyberattacks in IoT Spectrum Sensors

Device fingerprinting combined with Machine and Deep Learning (ML/DL) report promising performance when detecting cyberattacks targeting data managed by resource-constrained spectrum sensors. However, the amount of data needed to train models and the privacy concerns of such scenarios limit the applicability of centralized ML/DL-based approaches. Federated learning (FL) addresses these limitations by creating federated and privacy-preserving models. However, FL is vulnerable to malicious participants, and the impact of adversarial attacks on federated models detecting spectrum sensing data falsification (SSDF) attacks on spectrum sensors has not been studied. To address this challenge, the first contribution of this work is the creation of a novel dataset suitable for FL and modeling the behavior (usage of CPU, memory, or file system, among others) of resource-constrained spectrum sensors affected by different SSDF attacks. The second contribution is a pool of experiments analyzing and comparing the robustness of federated models according to i) three families of spectrum sensors, ii) eight SSDF attacks, iii) four scenarios dealing with unsupervised (anomaly detection) and supervised (binary classification) federated models, iv) up to 33% of malicious participants implementing data and model poisoning attacks, and v) four aggregation functions acting as anti-adversarial mechanisms to increase the models robustness.

Landscapes of Memory and Power: The Archaeology of a Forgotten Kingdom in Ethiopia

Unlike their northern counterparts, the kingdoms of southern Ethiopia have received little attention by archaeologists. Their relatively late emergence and absence of literacy may explain this lack of interest. However, they have much to offer to better understand the history not only of the Horn but also of the precolonial African state more generally. In this paper, the polities that developed in the southern Ethiopian highlands during the second millennium AD are briefly described and then one of them is explored in more detail: the kingdom of Anfillo. An archaeological and historical overview of the polity is provided based on two seasons of fieldwork. It is argued that in Anfillo, as in other southern Ethiopian polities, a fortified landscape materialized at the same time a persistent situation of conflict and the collective memory of the ruling classes, which used it as a mnemonic device to tell history and legitimize social divisions.