A Low-Power RRAM Memory Block for Embedded, Multi-Level Weight and Bias Storage in Artificial Neural Networks.
Ontology highlight
ABSTRACT:
SUBMITTER: Pechmann S
PROVIDER: S-EPMC8621881 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Bipolar resistance-switching materials allowing intermediate states of wide-varying resistance values hold the potential of drastically reduced power for non-volatile memory. To exploit this potential, we have introduced into a nanometallic resistance-random-access-memory (RRAM) device an asymmetric dynamic load, which can reliably lower switching power by orders of magnitude. The dynamic load is highly resistive during on-switching allowing access to the highly resistive intermediate states; during off-switching the load vanishes to enable switching at low voltage. This approach is entirely scalable and applicable to other bipolar RRAM with intermediate states. The projected power is 12 nW for a 100 × 100 nm(2) device and 500 pW for a 10 × 10 nm(2) device. The dynamic range of the load can be increased to allow power to be further decreased by taking advantage of the exponential decay of wave-function in a newly discovered nanometallic random material, reaching possibly 1 pW for a 10×10 nm(2) nanometallic RRAM device.
| S-EPMC3473363 | biostudies-literature
Project description:Monolayer transition metal dichalcogenides are considered to be promising candidates for flexible and transparent optoelectronics applications due to their direct bandgap and strong light-matter interactions. Although several monolayer-based photodetectors have been demonstrated, single-layered optical memory devices suitable for high-quality image sensing have received little attention. Here we report a concept for monolayer MoS2 optoelectronic memory devices using artificially-structured charge trap layers through the functionalization of the monolayer/dielectric interfaces, leading to localized electronic states that serve as a basis for electrically-induced charge trapping and optically-mediated charge release. Our devices exhibit excellent photo-responsive memory characteristics with a large linear dynamic range of ∼4,700 (73.4 dB) coupled with a low OFF-state current (<4 pA), and a long storage lifetime of over 104 s. In addition, the multi-level detection of up to 8 optical states is successfully demonstrated. These results represent a significant step toward the development of future monolayer optoelectronic memory devices.
| S-EPMC5376667 | biostudies-literature
Project description:Previous work supports an age-specific impairment for recognition memory of pairs of words and other stimuli. The present study tested the generalization of an associative deficit across word, name, and nonword stimulus types in younger and older adults. Participants completed associative and item memory tests in one of three stimulus conditions and made metacognitive ratings of perceptions of self-efficacy, task success ("postdictions"), strategy success, task effort, difficulty, fatigue, and stamina. Surprisingly, no support was found for an age-related associative deficit on any of the stimulus types. We analyzed our data further using a multilayer perceptron artificial neural network. The network was trained to classify individuals as younger or older and its hidden unit activities were examined to identify data patterns that distinguished younger from older participants. Analysis of hidden unit activities revealed that the network was able to correctly classify by identifying three different clusters of participants, with two qualitatively different groups of older individuals. One cluster of older individuals found the tasks to be relatively easy, they believed they had performed well, and their beliefs were accurate. The other cluster of older individuals found the tasks to be difficult, believed they were performing relatively poorly, yet their beliefs did not map accurately onto their performance. Crucially, data from the associative task were more useful for neural networks to discriminate between younger and older adults than data from the item task. This work underscores the importance of considering both individual and age differences as well as metacognitive responses in the context of associative memory paradigms.
| S-EPMC6668824 | biostudies-literature
Project description:Analogue memory-based deep neural networks provide energy-efficiency and per-area throughput gains relative to state-of-the-art digital counterparts such as graphics processing units. Recent advances focus largely on hardware-aware algorithmic training and improvements to circuits, architectures, and memory devices. Optimal translation of software-trained weights into analogue hardware weights-given the plethora of complex memory non-idealities-represents an equally important task. We report a generalised computational framework that automates the crafting of complex weight programming strategies to minimise accuracy degradations during inference, particularly over time. The framework is agnostic to network structure and generalises well across recurrent, convolutional, and transformer neural networks. As a highly flexible numerical heuristic, the approach accommodates arbitrary device-level complexity, making it potentially relevant for a variety of analogue memories. By quantifying the limit of achievable inference accuracy, it also enables analogue memory-based deep neural network accelerators to reach their full inference potential.
| S-EPMC9247051 | biostudies-literature
Project description:Large-scale solar energy production is still a great deal of obstruction due to the unpredictability of solar power. The intermittent, chaotic, and random quality of solar energy supply has to be dealt with by some comprehensive solar forecasting technologies. Despite forecasting for the long-term, it becomes much more essential to predict short-term forecasts in minutes or even seconds prior. Because key factors such as sudden movement of the clouds, instantaneous deviation of temperature in ambiance, the increased proportion of relative humidity and uncertainty in the wind velocities, haziness, and rains cause the undesired up and down ramping rates, thereby affecting the solar power generation to a greater extent. This paper aims to acknowledge the extended stellar forecasting algorithm using artificial neural network common sensical aspect. Three layered systems have been suggested, consisting of an input layer, hidden layer, and output layer feed-forward in conjunction with back propagation. A prior 5-min te output forecast fed to the input layer to reduce the error has been introduced to have a more precise forecast. Weather remains the most vital input for the ANN type of modeling. The forecasting errors might enhance considerably, thereby affecting the solar power supply relatively due to the variations in the solar irradiations and temperature on any forecasting day. Prior approximation of stellar radiations exhibits a small amount of qualm depending upon climatic conditions such as temperature, shading conditions, soiling effects, relative humidity, etc. All these environmental factors incorporate uncertainty regarding the prediction of the output parameter. In such a case, the approximation of PV output could be much more suitable than direct solar radiation. This paper uses Gradient Descent (GD) and Levenberg Maquarndt Artificial Neural Network (LM-ANN) techniques to apply to data obtained and recorded milliseconds from a 100 W solar panel. The essential purpose of this paper is to establish a time perspective with the greatest deal for the output forecast of small solar power utilities. It has been observed that 5 ms to 12 h time perspective gives the best short- to medium-term prediction for April. A case study has been done in the Peer Panjal region. The data collected for four months with various parameters have been applied randomly as input data using GD and LM type of artificial neural network compared to actual solar energy data. The proposed ANN based algorithm has been used for unswerving petite term forecasting. The model output has been presented in root mean square error and mean absolute percentage error. The results exhibit a improved concurrence between the forecasted and real models. The forecasting of solar energy and load variations assists in fulfilling the cost-effective aspects.
| S-EPMC10212924 | biostudies-literature
Project description:Resistive switching (RS) based on the formation and rupture of conductive filament (CF) is promising in novel memory and logic device applications. Understanding the physics of RS and the nature of CF is of utmost importance to control the performance, variability and reliability of resistive switching memory (RRAM). Here, the RESET switching of HfO2-based RRAM was statistically investigated in terms of the CF conductance evolution. The RESET usually combines an abrupt conductance drop with a progressive phase ending with the complete CF rupture. RESET1 and RESET2 events, corresponding to the initial and final phase of RESET, are found to be controlled by the voltage and power in the CF, respectively. A Monte Carlo simulator based on the thermal dissolution model of unipolar RESET reproduces all of the experimental observations. The results contribute to an improved physics-based understanding on the switching mechanisms and provide additional support to the thermal dissolution model.
| S-EPMC3796310 | biostudies-literature
Project description:Research exploring weight bias and weight bias internalisation (WBI) is grounded upon several core measures. This study aimed to evaluate whether operationalisations of these measures matched their conceptualisations in the literature. Using a 'closed card-sorting' methodology, participants sorted items from the most used measures into pre-defined categories, reflecting weight bias and non-weight bias. Findings indicated a high degree of congruence between WBI conceptualisations and operationalisations, however found less congruence between weight bias conceptualisations and operationalisations, with scale-items largely sorted into non-weight bias domains. Recommendations for scale modifications and developments are presented alongside a new amalgamated weight bias scale (AWBS).
| S-EPMC8323429 | biostudies-literature
Project description:A metal-free three component cyclization reaction with amidation is devised for direct synthesis of DFT-designed amido-phenazine derivative bearing noncovalent gluing interactions to fabricate organic nanomaterials. Composition-dependent organic nanoelectronics for nonvolatile memory devices are discovered using mixed phenazine-stearic acid (SA) nanomaterials. We discovered simultaneous two different types of nonmagnetic and non-moisture sensitive switching resistance properties of fabricated devices utilizing mixed organic nanomaterials: (a) sample-1(8:SA = 1:3) is initially off, turning on at a threshold, but it does not turn off again with the application of any voltage, and (b) sample-2 (8:SA = 3:1) is initially off, turning on at a sharp threshold and off again by reversing the polarity. No negative differential resistance is observed in either type. These samples have different device implementations: sample-1 is attractive for write-once-read-many-times memory devices, such as novel non-editable database, archival memory, electronic voting, radio frequency identification, sample-2 is useful for resistive-switching random access memory application.
| S-EPMC5645374 | biostudies-literature
Project description:If a three-dimensional physical electronic system emulating synapse networks could be built, that would be a significant step toward neuromorphic computing. However, the fabrication complexity of complementary metal-oxide-semiconductor architectures impedes the achievement of three-dimensional interconnectivity, high-device density, or flexibility. Here we report flexible three-dimensional artificial chemical synapse networks, in which two-terminal memristive devices, namely, electronic synapses (e-synapses), are connected by vertically stacking crossbar electrodes. The e-synapses resemble the key features of biological synapses: unilateral connection, long-term potentiation/depression, a spike-timing-dependent plasticity learning rule, paired-pulse facilitation, and ultralow-power consumption. The three-dimensional artificial synapse networks enable a direct emulation of correlated learning and trainable memory capability with strong tolerances to input faults and variations, which shows the feasibility of using them in futuristic electronic devices and can provide a physical platform for the realization of smart memories and machine learning and for operation of the complex algorithms involving hierarchical neural networks.High-density information storage calls for the development of modern electronics with multiple stacking architectures that increase the complexity of three-dimensional interconnectivity. Here, Wu et al. build a stacked yet flexible artificial synapse network using layer-by-layer solution processing.
| S-EPMC5622032 | biostudies-literature
Project description:Recently, artificial synapses involving an electrochemical reaction of Li-ion have been attributed to have remarkable synaptic properties. Three-terminal synaptic transistors utilizing Li-ion intercalation exhibits reliable synaptic characteristics by exploiting the advantage of non-distributed weight updates owing to stable ion migrations. However, the three-terminal configurations with large and complex structures impede the crossbar array implementation required for hardware neuromorphic systems. Meanwhile, achieving adequate synaptic performances through effective Li-ion intercalation in vertical two-terminal synaptic devices for array integration remains challenging. Here, two-terminal Au/LixCoO2/Pt artificial synapses are proposed with the potential for practical implementation of hardware neural networks. The Au/LixCoO2/Pt devices demonstrated extraordinary neuromorphic behaviors based on a progressive dearth of Li in LixCoO2 films. The intercalation and deintercalation of Li-ion inside the films are precisely controlled over the weight control spike, resulting in improved weight control functionality. Various types of synaptic plasticity were imitated and assessed in terms of key factors such as nonlinearity, symmetricity, and dynamic range. Notably, the LixCoO2-based neuromorphic system outperformed three-terminal synaptic transistors in simulations of convolutional neural networks and multilayer perceptrons due to the high linearity and low programming error. These impressive performances suggest the vertical two-terminal Au/LixCoO2/Pt artificial synapses as promising candidates for hardware neural networks.
| S-EPMC10030746 | biostudies-literature