RESUMO
In recent years, The combination of Attention mechanism and deep learning has a wide range of applications in the field of medical imaging. However, due to its complex computational processes, existing hardware architectures have high resource consumption or low accuracy, and deploying them efficiently to DNN accelerators is a challenge. This paper proposes an online-programmable Attention hardware architecture based on compute-in-memory (CIM) marco, which reduces the complexity of Attention in hardware and improves integration density, energy efficiency, and calculation accuracy. First, the Attention computation process is decomposed into multiple cascaded combinatorial matrix operations to reduce the complexity of its implementation on the hardware side; second, in order to reduce the influence of the non-ideal characteristics of the hardware, an online-programmable CIM architecture is designed to improve calculation accuracy by dynamically adjusting the weights; and lastly, it is verified that the proposed Attention hardware architecture can be applied for the inference of deep neural networks through Spice simulation. Based on the 100nm CMOS process, compared with the traditional Attention hardware architectures, the integrated density and energy efficiency are increased by at least 91.38 times, and latency and computing efficiency are improved by about 12.5 times.
RESUMO
Facial expression recognition has made a significant progress as a result of the advent of more and more convolutional neural networks (CNN). However, with the improvement of CNN, the models continues to get deeper and larger so as to a greater focus on the high-level features of the image and the low-level features tend to be lost. Because of the reason above, the dependence of low-level features between different areas of the face often cannot be summarized. In response to this problem, we propose a novel network based on the CNN model. To extract long-range dependencies of low-level features, multiple attention mechanisms has been introduced into the network. In this paper, the patch attention mechanism is designed to obtain the dependence between low-level features of facial expressions firstly. After fusion, the feature maps are input to the backbone network incorporating convolutional block attention module (CBAM) to enhance the feature extraction ability and improve the accuracy of facial expression recognition, and achieve competitive results on three datasets CK+ (98.10%), JAFFE (95.12%) and FER2013 (73.50%). Further, according to the PA Net designed in this paper, a hardware friendly implementation scheme is designed based on memristor crossbars, which is expected to provide a software and hardware co-design scheme for edge computing of personal and wearable electronic products.
RESUMO
Two types of multiweighted coupled memristive neural networks (CMNNs) with adaptive couplings are introduced in this article, and the fixed-time passivity (FXTP) and fixed-time synchronization (FXTS) of such networks are considered. First, under the developed adaptive scheme, a sufficient condition to guarantee the FXTP for multiweighted CMNNs with adaptive couplings is obtained. Second, the FXTP, fixed-time input-strict passivity and fixed-time output-strict passivity for multiweighted CMNNs with adaptive couplings and coupling delays are investigated by devising an appropriate state feedback controller. Third, applying the Lyapunov functional method, it establishes the FXTS criteria for the two kinds of networks presented. Finally, numerical examples are provided to demonstrate the effectiveness of the derived results.
RESUMO
In-sensor and near-sensor computing are becoming the next-generation computing paradigm for high-density and low-power sensory processing. To fulfil a high-density and efficient neuromorphic visual system with fully hierarchical emulation of the retina and visual cortex, emerging multimodal neuromorphic devices for multi-stage processing and a fully hardware-implemented system with versatile image processing functions are still lacking and highly desirable. Here we demonstrate an emerging multimodal-multifunctional resistive random-access memory (RRAM) device array based on modified silk fibroin protein (MSFP), exhibiting both optoelectronic RRAM (ORRAM) mode featured by unique negative and positive photoconductance memory and electrical RRAM (ERRAM) mode featured by analogue resistive switching. A full hardware implementation of the artificial visual system with versatile image processing functions is realised for the first time, including ORRAM mode array for the in-sensor image pre-processing (contrast enhancement, background denoising, feature extraction) and ERRAM mode array for near-sensor high-level image recognition, which hugely improves the integration density, and simply the circuit design and the fabrication and integration complexity.
RESUMO
BACKGROUND: Childhood maltreatment is considered as a robust predictor of depression. However, the underlying psychological and neurological mechanisms linking childhood maltreatment and depression remain poorly understood. Sufficient evidence demonstrates emotion dysregulation in individuals who have experienced childhood maltreatment, but it is unknown whether these changes represent vulnerability for depression. Here we speculated that decreased cognitive reappraisal and its corresponding neural basis might explain the relationship between childhood maltreatment and follow-up depression. METHODS: First, we investigated whether cognitive reappraisal can explain the relationship between childhood maltreatment and depression, with a cross-sectional (n = 657) behavioral sample. Then we recruit 38 maltreated participants and 27 controls to complete the cognitive reappraisal functional magnetic resonance imaging (fMRI) task. The between-group difference in brain activation and functional connectivity (FC) were tested using independent t-tests. Finally, we investigated the relationship between childhood maltreatment, task-based brain activity and depression. RESULTS: The behavior results suggested that cognitive reappraisal mediates the association between childhood maltreatment and depression. In addition, the maltreated group exhibited lower activation of orbitofrontal cortex (OFC) and higher FC of between the dorsolateral prefrontal cortex (DLPFC), posterior cingulate cortex (PCC), OFC, and amygdala during cognitive reappraisal, compared with healthy controls. Furthermore, the FC of DLPFC-amygdala mediates the association between childhood maltreatment and depression. CONCLUSION: In summary, childhood maltreatment is associated with inefficient cognitive reappraisal ability, manifesting as aberrant modulation of cortical areas on amygdala. These cognitive and neural deficits might explain the relationship between childhood maltreatment and risk of depression in later life.
Assuntos
Maus-Tratos Infantis , Depressão , Humanos , Criança , Depressão/psicologia , Estudos Transversais , Tonsila do Cerebelo/diagnóstico por imagem , Córtex Pré-Frontal , Mapeamento Encefálico , Imageamento por Ressonância Magnética , Cognição , Maus-Tratos Infantis/psicologia , Emoções/fisiologiaRESUMO
Artificial neural networks (ANNs) have gained considerable momentum in the past decade. Although at first the main task of the ANN paradigm was to tune the connection weights in fixed-architecture networks, there has recently been growing interest in evolving network architectures toward the goal of creating artificial general intelligence. Lagging behind this trend, current ANN hardware struggles for a balance between flexibility and efficiency but cannot achieve both. Here, we report on a novel approach for the on-demand generation of complex networks within a single memristor where multiple virtual nodes are created by time multiplexing and the non-trivial topological features, such as small-worldness, are generated by exploiting device dynamics with intrinsic cycle-to-cycle variability. When used for reservoir computing, memristive complex networks can achieve a noticeable increase in memory capacity a and respectable performance boost compared to conventional reservoirs trivially implemented as fully connected networks. This work expands the functionality of memristors for ANN computing.
RESUMO
Echo state network (ESN) has gained wide acceptance in the field of time series prediction, relying on sufficiently complex reservoir connections to remember the historical features of the data and using these features to obtain the outputs by a simple linear readout. However, the randomness of its input and reservoir connections pose negative impacts on the prediction performance and performance stability of the models, the complexity of reservoir connections brings high time consumption during network computing, and the presence of randomness and complexity makes the hardware implementation of the ESN difficult. In response, we propose a double-cycle ESN (DCESN) based on the Li-ESN model, which has fixed weights to improve prediction performance and performance stability and simpler reservoir connections compared to the classical ESN to reduce the time consumption. The existence of both greatly reduces the difficulty of hardware implementation of the ESN and provides many conveniences for the future application of the ESN. Experimental results on many widely used time series datasets show that the DCESN has comparable or even better prediction performance than the ESN and good robustness against noise and parameter fluctuations.
RESUMO
In the field of second language acquisition, overshadowing and blocking by cue competition effects in classical conditioning affect the learning and expression of human cognitive associations. In this work, a memristive neural network circuit based on neurobiological mechanisms is proposed, which consists of synapse module, neuron module, and control module. In particular, the designed network introduces an inhibitory interneuron to divide memristive synapses into excitatory and inhibitory memristive synapses, so as to mimic synaptic plasticity better. In addition, the proposed circuit can implement six functions of second language acquisition conditioning, including learning, overshadowing, blocking, recovery from overshadowing, recovery from blocking, and long-term effect of overshadowing over time leading to blocking. Overshadowing, which denotes that the more salient stimulus overshadows the learning of the less salient stimulus when two stimuli differ in salience, reduces the associative strength acquired by the less salient stimulus. Blocking, which indicates that pretraining on one stimulus blocks learning about a second stimulus, inhibits the associative strength acquired by a second stimulus. Finally, the correctness and effectiveness of implementing functions mentioned above are verified by the simulation results in PSPICE. Through further research, the proposed circuit is applied to bionic devices such as social robots or educational robots, which can address language and cognitive disorders via assisted learning and training.
RESUMO
With the introduction of neuron coverage as a testing criterion for deep neural networks (DNNs), covering more neurons to detect more internal logic of DNNs became the main goal of many research studies. While some works had made progress, some new challenges for testing methods based on neuron coverage had been proposed, mainly as establishing better neuron selection and activation strategies influenced not only obtaining higher neuron coverage, but also more testing efficiency, validating testing results automatically, labeling generated test cases to extricate manual work, and so on. In this article, we put forward Test4Deep, an effective white-box testing DNN approach based on neuron coverage. It is based on a differential testing framework to automatically verify inconsistent DNNs' behavior. We designed a strategy that can track inactive neurons and constantly triggered them in each iteration to maximize neuron coverage. Furthermore, we devised an optimization function that guided the DNN under testing to deviate predictions between the original input and generated test data and dominated unobservable generation perturbations to avoid manually checking test oracles. We conducted comparative experiments with two state-of-the-art white-box testing methods DLFuzz and DeepXplore. Empirical results on three popular datasets with nine DNNs demonstrated that compared to DLFuzz and DeepXplore, Test4Deep, on average, exceeded by 32.87% and 35.69% in neuron coverage, while reducing 58.37% and 53.24% testing time, respectively. In the meantime, Test4Deep also produced 58.37% and 53.24% more test cases with 23.81% and 98.40% fewer perturbations. Even compared with the two highest neuron coverage strategies of DLFuzz, Test4Deep still enhanced neuron coverage by 4.34% and 23.23% and achieved 94.48% and 85.67% higher generation time efficiency. Furthermore, Test4Deep could improve the accuracy and robustness of DNNs by merging generated test cases and retraining.
Assuntos
Redes Neurais de Computação , NeurôniosRESUMO
Memristor-based Pavlov associative memory circuit presented today only realizes the simple condition reflex process. The secondary condition reflex endows the simple condition reflex process with more bionic, but it is only demonstrated in design and involves the large number of redundant circuits. A FeO x -based memristor exhibits an evolution process from battery-like capacitance (BLC) state to resistive switching (RS) memory as the I-V sweeping increase. The BLC is triggered by the active metal ion and hydroxide ion originated from water molecule splitting at different interfaces, while the RS memory behavior is dominated by the diffusion and migration of ion in the FeO x switching function layer. The evolution processes share the nearly same biophysical mechanism with the second-order conditioning. It enables a hardware-implemented second-order associative memory circuit to be feasible and simple. This work provides a novel path to realize the associative memory circuit with the second-order conditioning at hardware level.
RESUMO
A memristor with Au/polyimide (PI)/Au structure is prepared by magnetron sputtering to investigate the multiphotoconductance resistive switching (RS) memory behavior. The PI-based memristor presents stable bipolar RS memory and is sensitive to visible light. Four discrete conductance states in both high-resistance state (HRS) and low-resistance state (LRS) are obtained when illuminating by 365, 550, 590, and 780 nm light. Electron trapping and detrapping from the defects distributed at interfaces and the PI switching layer are responsible for the observed RS memory behavior. The enhanced trapping and detrapping process by light illumination is responsible for the multiconductance states. This work provides the possibility for further development of neuromorphic vision sensors using an organic semiconductor-based memristor.
RESUMO
Unipolar resistive switching (URS) behavior, known as the SET and RESET operating in a single voltage sweep direction, has shown great potential in the simplification of the peripheral circuit. The URS memristor always involves complicated interfacial engineering and structural design. In this work, a reliable URS behavior is realized using a simple Ag/HfOx/Pt memristor structure. The memristor displays a retention time of >104 s, an ON/OFF ratio of >103, and a good operation voltage. Synergy and competition between the Ag conductive filament formed by redox reaction and the migration of an oxygen vacancy are responsible for the observed URS. By comparison, a 35% power consumption is reduced during the logical operation from 0 to 1 to 0. The operation strategy is demonstrated by exhibiting the ACSII code of the capital letter denoted by eight logic states. This work provides a low-power concept for ultrahigh data storage using the URS memristor.
RESUMO
The signal transmission mechanism of the Resistor-Capacitor (RC) circuit is similar to the intracellular and extracellular signal propagating mechanism of the neuron. Thus, the RC circuit can be utilized as the circuit model of the neuron cell membrane. However, resistors are electronic components with the fixed-resistance and have no memory properties. A memristor is a promising neuro-morphological electronic device with nonvolatile, switching, and nonlinear characteristics. First of all, we consider replacing the resistor in the RC neuron circuit with a memristor, which is named the Memristor-Capacitor (MC) circuit, then the MC neuron model is constructed. We compare the charging and discharging processes between the RC and MC neuron circuits. Secondly, two models are compared under the different external stimuli. Finally, the synchronous and asynchronous activities of the RC and MC neuron circuits are performed. Extensive experimental results suggest that the charging and discharging speed of the MC neuron circuit is faster than that of the RC neuron circuit. Given sufficient time and proper external stimuli, the RC and MC neuron circuits can produce the action potentials. The synchronous and asynchronous phenomena in the two neuron circuits reproduce nonlinear dynamic behaviors of the biological neurons.
RESUMO
Stressful life events are significant risk factors for depression, and increases in depressive symptoms have been observed during the COVID-19 pandemic. The aim of this study is to explore the neural makers for individuals' depression during COVID-19, using connectome-based predictive modeling (CPM). Then we tested whether these neural markers could be used to identify groups at high/low risk for depression with a longitudinal dataset. The results suggested that the high-risk group demonstrated a higher level and increment of depression during the pandemic, as compared to the low-risk group. Furthermore, a support vector machine (SVM) algorithm was used to discriminate major depression disorder patients and healthy controls, using neural features defined by CPM. The results confirmed the CPM's ability for capturing the depression-related patterns with individuals' resting-state functional connectivity signature. The exploration for the anatomy of these functional connectivity features emphasized the role of an emotion-regulation circuit and an interoception circuit in the neuropathology of depression. In summary, the present study augments current understanding of potential pathological mechanisms underlying depression during an acute and unpredictable life-threatening event and suggests that resting-state functional connectivity may provide potential effective neural markers for identifying susceptible populations. (PsycInfo Database Record (c) 2022 APA, all rights reserved).
Assuntos
COVID-19 , Conectoma , Transtorno Depressivo Maior , Encéfalo/diagnóstico por imagem , Conectoma/métodos , Depressão , Humanos , Individualidade , Imageamento por Ressonância Magnética/métodos , PandemiasRESUMO
This study aimed to analyze the influence of artificial intelligence (AI) reconstruction algorithm on computed tomography (CT) images and the application of CT image analysis in the recovery of knee anterior cruciate ligament (ACL) sports injuries. A total of 90 patients with knee trauma were selected for enhanced CT scanning and randomly divided into three groups. Group A used the filtered back projection (FBP) reconstruction algorithm, and the tube voltage was set to 120 kV during CT scanning. Group B used the iDose4 reconstruction algorithm, and the tube voltage was set to 120 kV during CT scanning. In group C, the iDose4 reconstruction algorithm was used, and the tube voltage was set to 100 kV during CT scanning. The noise, signal-to-noise ratio (SNR), carrier-to-noise ratio (CNR), CT dose index volume (CTDI), dose length product (DLP), and effective radiation dose (ED) of the three groups of CT images were compared. The results showed that the noise of groups B and C was smaller than that of group A (P < 0.05), and the SNR and CNR of groups B and C were higher than those of group A. The images of patients in group A with the FBP reconstruction algorithm were noisy, and the boundaries were not clear. The noise of the images obtained by the iDose4 reconstruction algorithm in groups B and C was improved, and the image resolution was also higher. The agreement between arthroscopy and CT scan results was 96%. Therefore, the iterative reconstruction algorithm of iDose4 can improve the image quality. It was of important value in the diagnosis of knee ACL sports injury.
Assuntos
Traumatismos em Atletas , Algoritmos , Ligamento Cruzado Anterior , Inteligência Artificial , Traumatismos em Atletas/diagnóstico por imagem , Humanos , Doses de Radiação , Tomografia Computadorizada por Raios X/métodosRESUMO
The objective of this research was to investigate the application value of deep learning-based computed tomography (CT) images in the diagnosis of liver tumors. Fifty-eight patients with liver tumors were selected, and their CT images were segmented using a convolutional neural network (CNN) algorithm. The segmentation results were quantitatively evaluated using the Dice similarity coefficient (DSC), precision, and recall. All the patients were examined and diagnosed by CT enhanced delayed scan technique, and the CT scan results were compared with the pathological findings. The results showed that the DSC, precision, and recall of the CNN algorithm reached 0.987, 0.967, and 0.954, respectively. The images segmented by the CNN were clearer. The diagnostic result of the examination on 56 cases by CT enhanced delay scanning was consistent with that of pathological diagnosis. According to the result of pathological diagnosis, there were 6 cases with hepatic cyst, 9 with hepatic hemangioma, 12 cases with liver metastasis, 10 cases with hepatoblastoma, 3 cases with focal nodular hyperplasia, and 18 cases with primary liver cancer. The result of CT enhanced delay scanning on 58 patients was consistent with that of pathological diagnosis, and the total diagnostic coincidence rate reached 96.55%. In conclusion, the CNN algorithm can perform accurate and efficient segmentation, with high resolution, providing a more scientific basis for the segmentation of liver tumors in CT images. CT enhanced scanning technology has a good effect on the diagnosis and differentiation of liver tumor patients, with high diagnostic coincidence rate. It has important value for the diagnosis of liver tumor and is worthy of clinical application.
Assuntos
Aprendizado Profundo , Neoplasias Hepáticas , Humanos , Processamento de Imagem Assistida por Computador/métodos , Neoplasias Hepáticas/diagnóstico por imagem , Redes Neurais de Computação , Tomografia Computadorizada por Raios X/métodosRESUMO
The Izhikevich (IZH) spiking neuron model can display spiking and bursting behaviors of neurons. Based on the switching property and bio-plausibility of the memristor, the memristive Izhikevich (MIZH) spiking neuron model is built. Firstly, the MIZH spiking model is introduced and used to generate 23 spiking patterns. We compare the 23 spiking patterns produced by the IZH and MIZH spiking models. Secondly, the MIZH spiking model actively reproduces various neuronal behaviors, including the excitatory cortical neurons, the inhibitory cortical neurons, and other cortical neurons. Finally, the collective dynamic activities of the MIZH neuronal network are performed, and the MIZH oscillatory network is constructed. Experimental results illustrate that the constructed MIZH spiking neuron model performs high firing frequency and good frequency adaptation. The model can easily simulate various spiking and bursting patterns of distinct neurons in the brain. The MIZH neuronal network realizes the synchronous and asynchronous collective behaviors. The MIZH oscillatory network can memorize and retrieve the information patterns correctly and efficiently with high retrieval accuracy.
RESUMO
The leaky integrate-and-fire (LIF) spiking model can successively mimic the firing patterns and information propagation of a biological neuron. It has been applied in neural networks, cognitive computing, and brain-inspired computing. Due to the resistance variability and the natural storage capacity of the memristor, the LIF spiking model with a memristor (MLIF) is presented in this article to simulate the function and working mode of neurons in biological systems. First, the comparison between the MLIF spiking model and the LIF spiking model is conducted. Second, it is experimentally shown that a single memristor could mimic the function of the integration and filtering of the dendrite and emulate the function of the integration and firing of the soma. Finally, the feasibility of the proposed MLIF spiking model is verified by the generation and recognition of Morse code. The experimental results indicate that the presented MLIF model efficiently performs good biological frequency adaptation, high firing frequency, and rich spiking patterns. A memristor can be used as the dendrite and the soma, and the MLIF spiking model can emulate the axon. The constructed single neuron can efficiently complete the generation and propagation of firing patterns.
RESUMO
This study aimed to discuss magnetic resonance imaging (MRI) based on artificial intelligence (AI) algorithm to evaluate the effect of radiation synovectomy for hemophilic arthropathy (HA). MRI based on the Canny algorithm was applied and compared with conventional MRI to evaluate its application effects according to the PSNR and SSIM. Sixty patients diagnosed with HA were selected as the research subjects. According to the detection method, the patients were divided into group A (pathological detection after radiation synovectomy), group B (conventional MRI detection), and group C (MRI detection based on the Canny algorithm). The application value of MRI based on the Canny algorithm was judged by comparing the differences between the two detection methods and pathological results. The results displayed that the reconstruction effect of the Canny algorithm was remarkably better than that of the traditional algorithm regarding the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), which showed a clearer synovial contour. The results of the IPSG score of joint effusion and hemorrhage showed that there was a difference in the detection rate of joints between conventional MRI and pathological results on the score of 1 and 2 (P < 0.05); and there was no significant difference between the MRI and pathological results based on the Canny algorithm (P > 0.05). The results of the IPSG score of synovial hyperplasia showed that the detection rate of conventional MRI was different from pathological results on the score of 1 and 2 (P < 0.05); and there was no significant difference between the MRI and pathological results based on the Canny algorithm (P > 0.05). The results of the IPSG score of hemosiderin deposition showed that the detection rate of conventional MRI was different from the pathological results on the score of 1 and 2 (P < 0.05); and there was no significant difference between the MRI and pathological results based on the Canny algorithm (P > 0.05). The synovial volume of patients after surgery was reduced compared with that before surgery. One-factor variance was used to analyze the clinical hemorrhage frequency before and after surgery, and the results showed that the differences were statistically significant (P < 0.05). Therefore, MRI on account of AI algorithm made it easier to detect synovial contour, which was helpful to evaluate the efficacy of polygenic risk scores (PRS) surgery in HA patients. MRI based on the Canny algorithm had less differences between the score of hemophilic arthropathy and pathological results, which could replace conventional MRI examination and have clinical application value.
Assuntos
Artropatias , Sinovectomia , Algoritmos , Inteligência Artificial , Humanos , Artropatias/diagnóstico por imagem , Artropatias/cirurgia , Imageamento por Ressonância MagnéticaRESUMO
Over the past few decades, the research of dissipative chaotic systems has yielded many achievements in both theory and application. However, attractors in dissipative systems are easily reconstructed by the attacker, which leads to information security problems. Compared with dissipative systems, conservative ones can effectively avoid these reconstructing attacks due to the absence of attractors. Therefore, conservative systems have advantages in chaos-based applications. Currently, there are still relatively few studies on conservative systems. For this purpose, based on the simplest memristor circuit in this paper, a non-Hamiltonian 3D conservative system without equilibria is proposed. The phase volume conservatism is analyzed by calculating the divergence of the system. Furthermore, a Kolmogorov-type transformation suggests that the Hamiltonian energy is not conservative. The most prominent property in the conservative system is that it exhibits quasi-periodic 3D tori with heterogeneous coexisting and different amplitude rescaling trajectories triggered by initial values. In addition, the results of Spectral Entropy analysis and NIST test show that the system can produce pseudo-random numbers with high randomness. To the best of our knowledge, there is no 3D conservative system with such complex dynamics, especially in a memristive conservative system. Finally, the analog circuit of the system is designed and implemented to test its feasibility as well.