Search

Spiking Neural Networks (SNNs) have become one of the most prominent next-generation computational models owing to their biological plausibility, low power consumption, and the potential for neuromorphic hardware implementation. Among the various methods for obtaining available SNNs, converting Artificial Neural Networks (ANNs) into SNNs is the most cost-effective approach. The early challenges in ANN-to-SNN conversion work revolved around the susceptibility of converted SNNs to conversion errors. Some recent endeavors have attempted to mitigate these conversion errors by altering the original ANNs. Despite their ability to enhance the accuracy of SNNs, these methods lack generality and cannot be directly applied to convert the majority of existing ANNs. In this paper, we present a framework named DNISNM for converting ANN to SNN, with the aim of addressing conversion errors arising from differences in the discreteness and asynchrony of network transmission between ANN and SNN. The DNISNM consists of two mechanisms, Data-based Neuronal Initialization (DNI) and Signed Neuron with Memory (SNM), designed to respectively address errors stemming from discreteness and asynchrony disparities. This framework requires no additional modifications to the original ANN and can result in SNNs with improved accuracy performance, simultaneously ensuring universality, high precision, and low inference latency. We verify it experimentally on challenging object recognition datasets, including CIFAR10, CIFAR100, and ImageNet-1k. Experimental results show that the SNN converted by our framework has very high accuracy even at extremely low latency., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 Elsevier Ltd. All rights reserved.)

Episodic memory is fundamental to the brain's cognitive function, but how neuronal activity is temporally organized during its encoding and retrieval is still unknown. In this article, combining hippocampus structure with a spiking neural network (SNN), a new bionic spiking temporal memory (BSTM) model is proposed to explore the encoding, formation, and retrieval of episodic memory. For encoding episodic memory, the spike-timing-dependent-plasticity (STDP) learning algorithm and a proposed minicolumn selection algorithm are used to encode each input item into several active minicolumns. For the formation of episodic memory, a sequential memory algorithm is proposed to store the contexts between items. For retrieval of episodic memory, the local retrieval algorithm and the global retrieval algorithm are proposed to retrieve sequence information, achieving multisentence prediction and multitime step prediction. All functions of BSTM are based on bionic spiking neurons, which have biological characteristics including columnar and dendritic structures, firing and receiving spikes, and delaying transmission. To test the performance of the BSTM model, the Children's Book Test (CBT) data set was used to conduct a series of experiments under different settings, including changing the number of minicolumns, neurons and sequences, modifying sequence items, etc. Compared to other sequence memory algorithms, the experimental results show that the proposed BSTM achieves higher accuracy and better robustness.

Spiking neurons are becoming increasingly popular owing to their biological plausibility and promising computational properties. Unlike traditional rate-based neural models, spiking neurons encode information in the temporal patterns of the transmitted spike trains, which makes them more suitable for processing spatiotemporal information. One of the fundamental computations of spiking neurons is to transform streams of input spike trains into precisely timed firing activity. However, the existing learning methods, used to realize such computation, often result in relatively low accuracy performance and poor robustness to noise. In order to address these limitations, we propose a novel highly effective and robust membrane potential-driven supervised learning (MemPo-Learn) method, which enables the trained neurons to generate desired spike trains with higher precision, higher efficiency, and better noise robustness than the current state-of-the-art spiking neuron learning methods. While the traditional spike-driven learning methods use an error function based on the difference between the actual and desired output spike trains, the proposed MemPo-Learn method employs an error function based on the difference between the output neuron membrane potential and its firing threshold. The efficiency of the proposed learning method is further improved through the introduction of an adaptive strategy, called skip scan training strategy, that selectively identifies the time steps when to apply weight adjustment. The proposed strategy enables the MemPo-Learn method to effectively and efficiently learn the desired output spike train even when much smaller time steps are used. In addition, the learning rule of MemPo-Learn is improved further to help mitigate the impact of the input noise on the timing accuracy and reliability of the neuron firing dynamics. The proposed learning method is thoroughly evaluated on synthetic data and is further demonstrated on real-world classification tasks. Experimental results show that the proposed method can achieve high learning accuracy with a significant improvement in learning time and better robustness to different types of noise.

Sec1/Munc18-family (SM) proteins are required for SNARE-mediated membrane fusion, but their mechanism(s) of action remain controversial. Using single-molecule force spectroscopy, we found that the SM protein Munc18-1 catalyzes step-wise zippering of three synaptic SNAREs (syntaxin, VAMP2, and SNAP-25) into a four-helix bundle. Catalysis requires formation of an intermediate template complex in which Munc18-1 juxtaposes the N-terminal regions of the SNARE motifs of syntaxin and VAMP2, while keeping their C-terminal regions separated. SNAP-25 binds the templated SNAREs to induce full SNARE zippering. Munc18-1 mutations modulate the stability of the template complex in a manner consistent with their effects on membrane fusion, indicating that chaperoned SNARE assembly is essential for exocytosis. Two other SM proteins, Munc18-3 and Vps33, similarly chaperone SNARE assembly via a template complex, suggesting that SM protein mechanism is conserved., Competing Interests: JJ, MH, SP, RB, YX, HQ, YX, YW, HJ, TE, FH, YZ No competing interests declared, (© 2018, Jiao et al.)

A better understanding is needed of how glutamate metabolism is affected in mesial temporal lobe epilepsy (MTLE). Here we investigated glial-neuronal metabolism in the chronic phase of the kainate (KA) model of MTLE. Thirteen weeks following systemic KA, rats were injected i.p. with [1-(13)C]glucose. Brain extracts from hippocampal formation, entorhinal cortex, and neocortex, were analyzed by (13)C and (1)H magnetic resonance spectroscopy to quantify (13)C labeling and concentrations of metabolites, respectively. The amount and (13)C labeling of glutamate were reduced in the hippocampal formation and entorhinal cortex of epileptic rats. Together with the decreased concentration of NAA, these results indicate neuronal loss. Additionally, mitochondrial dysfunction was detected in surviving glutamatergic neurons in the hippocampal formation. In entorhinal cortex glutamine labeling and concentration were unchanged despite the reduced glutamate content and label, possibly due to decreased oxidative metabolism and conserved flux of glutamate through glutamine synthetase in astrocytes. This mechanism was not operative in the hippocampal formation, where glutamine labeling was decreased. In neocortex labeling and concentration of GABA were increased in epileptic rats, possibly representing a compensatory mechanism. The changes in the hippocampus might be of pathophysiological importance and merit further studies aiming at resolving metabolic causes and consequences of MTLE.

Pyruvate recycling is a pathway for complete oxidation of glutamate. The cellular location and the physiological significance of such recycling has been debated during the last decade. The present study was aimed at elucidating whether recycling takes place in neuron-enriched cultures of dissociated cerebella, consisting mainly of glutamatergic granule cells, some GABAergic neurons, and few astrocytes. These cultures and cultures of astrocytes from cerebellum were incubated in medium containing [U-(13)C]glutamate, and cell extracts were analyzed by gas chromatography and mass spectrometry. Additionally, in the case of the neuron-enriched cultures, a magnetic resonance (MR) spectrum was obtained. It could be shown that the atom percentage excess of the isotopomer representing pyruvate recycling in glutamate (M + 4) was similar for astrocytes and neuron-enriched cultures. However, the latter showed more recycling in glutamine (synthesized in the small fraction of astrocytes) than the pure astrocyte cultures, whereas the reverse was the case for aspartate. In fact, the atom percentage excess of the isotopomer representing pyruvate recycling in glutamine was slightly but significantly higher than that in glutamate in the neuron-enriched cultures. It can be concluded that pyruvate recycling is clearly present in neurons, and this was verified by MR spectroscopy.

Glutamate neurotoxicity is implicated in most neurodegenerative diseases, and in the present study the long-term effects of the glutamate agonist kainic acid (KA) on cerebellar neurons are investigated. Primary cell cultures, mainly consisting of glutamatergic granule neurons, were cultured in medium containing 0.05 or 0.50mM KA for 7 days and subsequently incubated in medium containing [U-13C]glutamate or [U-13C]glutamine. The amount of protein and number of cells were greatly reduced in cultures exposed to 0.50 mM KA compared to those exposed to 0.05 mM KA. Glutamine consumption was not affected by KA concentration, whereas that of glutamate was decreased by high KA, confirming reduction in glutamate transport reported earlier. Neurons cultured with 0.50 mM KA and incubated with glutamate contained decreased amounts of glutamate, aspartate and GABA compared to those cultured with 0.05 mM KA. Incubation of cells exposed to 0.50 mM KA with glutamine led to an increased amount of glutamate compared to cells exposed to 0.05 mM KA, whereas the intracellular amounts of aspartate and GABA remained unaffected by KA concentration. Furthermore, mitochondrial metabolism of alpha-[U-13C]ketoglutarate derived from [U-13C]glutamate and [U-13C]glutamine was significantly reduced by 0.50 mM KA. The results presented illustrate differential vulnerability to KA and can only be understood in terms of inter- and intracellular compartmentation.

The glutamate-glutamine cycle is thought to be of paramount importance in the mature brain; however, its significance is likely to vary with regional differences in distance between astrocyte and synapse. The present study is aimed at evaluating the role of this cycle in cultures of cerebellar neurons, mainly consisting of glutamatergic granule cells. Cells were incubated in medium containing [U-13C]glutamate or [U-13C]glutamine in the presence and absence of unlabeled glutamine and glutamate, respectively. Cell extracts and media were analyzed using high-performance liquid chromatography (HPLC) and gas chromatography combined with mass spectrometry (GC/MS). Both [U-13C]glutamate and [U-13C]glutamine were shown to be excellent precursors for synthesis of neuroactive amino acids and tricarboxylic acid (TCA) cycle intermediates. Labeling from [U-13C]glutamate was higher than that from [U-13C]glutamine in all metabolites measured. The presence of [U-13C]glutamate plus unlabeled glutamine in the experimental medium led to labeling very similar to that from [U-13C]glutamate alone. However, incubation in medium containing [U-13C]glutamine in the presence of unlabeled glutamate almost abolished labeling of metabolites. Thus, it could be shown that glutamate is the preferred substrate for intermediary metabolism in cerebellar neurons. Label distribution indicating TCA cycle activity showed more prominent cycling from [U-13C]glutamine than from [U-13C]glutamate. Labeling of succinate was lower than that of the other TCA cycle intermediates, indicating an active role of the gamma-amino butyric acid shunt in these cultures. It can be concluded that the cerebellar neurons rely more on reuptake of glutamate than supply of glutamine from astrocytes for glutamate homeostasis.

Cultures of dissociated cerebella from 7-day-old mice were maintained in vitro for 1-13 days. GABA biosynthesis and degradation were studied during development in culture and pharmacological agents were used to identify the enzymes involved. The amount of GABA increased, whereas that of glutamate was unchanged during the first 5 days and both decreased thereafter. The presence of aminooxyacetic acid (AOAA, 10 microM) which inhibits transaminases and other pyridoxal phosphate dependent enzymes including GABA-transaminase (GABA-T), in the culture medium caused an increase in the intracellular amount of GABA and a decrease in glutamate. The GABA content was also increased following exposure to the specific GABA-T inhibitor gamma-vinyl GABA. From day 6 in culture (day 4 when cultured in the presence of AOAA) GABA levels in the medium were increased compared to that in medium from 1-day-old cultures. Synthesis of GABA during the first 3 days was demonstrated by the finding that incubation with either [1-(13)C]glucose or [U-(13)C]glutamine led to formation of labeled GABA. Synthesis of GABA after 1 week in culture, when the enzymatic machinery is considered to be at a more differentiated level, was shown by labeling from [U-(13)C]glutamine added on day 7. Altogether the findings show continuous GABA synthesis and degradation throughout the culture period in the cerebellar neurons. At 10 microM AOAA, GABA synthesis from [U-(13)C]glutamine was not affected, indicating that transaminases are not involved in GABA synthesis and thus excluding the putrescine pathway. At a concentration of 5 mM AOAA GABA labeling was, however, abolished, showing that glutamate decarboxylase, which is inhibited at this level of AOAA, is responsible for GABA synthesis in the cerebellar cultures. In conclusion, the present study shows that GABA synthesis is taking place via GAD in a subpopulation of the cerebellar neurons, throughout the culture period.

Objective: To examine the effect of Buyanghuanwu decoction (BYHWD) in inducing nerve proliferation in rats with sequelae of ischemic stroke., Methods: A rat model of ischemic stroke sequelae was established by means of craniectomy in which the right common carotid artery was ligated with 4-0 silk thread followed by cauterization of the right middle cerebral artery. Programmed electric shock was administered 24 h after the onset of ischemic stroke for 2 h daily for 20 consecutive days. The rats in sham operation group were not subjected to ligation of the right common carotid artery or right middle cerebral artery occlusion. The rats in the treatment groups were given oral BYHWD for 15 consecutive days. All the rats received repeated intraperitoneal injections of the cell proliferation-specific marker 5-bromodeoxyuridine (BrdU), and the intake of BrdU in the cerebral tissues was determined by immunohistochemistry., Results: The number of BrdU-immunoreactive cells in the cerebral tissues of BYHWD-treated rats was significantly greater than that in the untreated model group., Conclusion: BYHWD can promote nerve proliferation in rats with ischemic stroke sequelae.

Glutamate exists in a vesicular as well as a cytoplasmic pool and is metabolically closely related to the tricarboxylic acid (TCA) cycle. Glutamate released during neuronal activity is most likely to a large extent accumulated by astrocytes surrounding the synapse. A compensatory flux from astrocytes to neurons of suitable precursors is obligatory as neurons are incapable of performing a net synthesis of glutamate from glucose. Glutamine appears to play a major role in this context. Employing cultured cerebellar granule cells, as a model system for glutamatergic neurons, details of the biosynthetic machinery have been investigated during depolarizing conditions inducing vesicular release. [U-13C]Glucose and [U-13C]glutamine were used as labeled precursors for monitoring metabolic pathways by nuclear magnetic resonance (NMR) spectroscopy and liquid chromatography-mass spectrometry (LC-MS) technologies. To characterize release mechanisms and influence of glutamate transporters on maintenance of homeostasis in the glutamatergic synapse, a quantification was performed by HPLC analysis of the amounts of glutamate and aspartate released in response to depolarization by potassium (55 mM) in the absence and presence of DL-threo-beta-benzyloxyaspartate (TBOA) and in response to L-trans-pyrrolidine-2,4-dicarboxylate (t-2,4-PDC), a substrate for the glutamate transporter. Based on labeling patterns of glutamate the biosynthesis of the intracellular pool of glutamate from glutamine was found to involve the TCA cycle to a considerable extent (approximately 50%). Due to the mitochondrial localization of PAG this is unlikely only to reflect amino acid exchange via the cytosolic aspartate aminotransferase reaction. The involvement of the TCA cycle was significantly lower in the synthesis of the released vesicular pool of glutamate. However, in the presence of TBOA, inhibiting glutamate uptake, the difference between the intracellular and the vesicular pool with regard to the extent of involvement of the TCA cycle in glutamate synthesis from glutamine was eliminated. Surprisingly, the intracellular pool of glutamate was decreased after repetitive release from the vesicular pool in the presence of TBOA indicating that neuronal reuptake of released glutamate is involved in the maintenance of the neurotransmitter pool and that 0.5 mM glutamine exogenously supplied is inadequate to sustain this pool.

Culturing mouse cerebellar neurones (predominantly glutamatergic) in the presence of [1-(13)C]glucose for 7 days resulted in a surprisingly extensive labelling of the inhibitory neurotransmitter GABA, the average content and labelling of which were 20 +/- 4 nmol/mg protein and 20 +/- 4%, respectively. Cultures of neocortical neurones (predominantly GABAergic) had under similar conditions a GABA content and labelling of 32 +/- 2 nmol/mg protein and 21 +/- 2%. The cerebellar cultures contained only 6% glutamate decarboxylase (GAD)-positive neurones when immunolabelled using a GAD67 antibody, while a dense network of neurones in the neocortical cultures stained positively for GAD67. Exposure of the cerebellar cultures to 50 microm kainic acid (KA) which is known to eliminate vesicular release of GABA, only marginally affected GABA labelling and cellular content. Likewise this treatment had no effect on the number of GAD67-positive neurones but a massive punctate immunostaining observed in control cultures was essentially eliminated. Increasing the KA concentration to 0.5 mm in the culture medium for 7 days led to a reduction of GABA labelling and content compared to cerebellar cultures not exposed to KA. Although it is likely that this large capacity for GABA synthesis resides in the relatively few GAD-positive neurones, it seems unlikely that they could account for the large average GABA content in the cultures. Therefore it must be concluded that the newly synthesized GABA is redistributed among the majority of the cells in these cultures, i.e. the glutamatergic neurones.

Limbic seizures were induced in rats by intraperitoneal injection of the glutamate receptor agonist kainic acid, followed, 24h later by injection of [1-13C]glucose and [1,2-13C]acetate. Analyses of forebrain extracts were performed using 13C magnetic resonance spectroscopy and HPLC. A significant increase in label derived from [1,2-13C]acetate was observed in glutamine and glutamate. Label in most metabolites derived from [1-13C]glucose was unchanged, however, a decrease was observed in [2-13C]GABA, possibly due to reduced GABA release, 24h after kainic acid injection. It should be noted that only astrocytes are able to utilize acetate as a substrate efficiently, whereas acetyl CoA derived from glucose is metabolized predominantly in the neuronal tricarboxylic acid cycle. No significant differences were found in total amounts of amino acids between the two groups. Thus, these results indicate that turnover of metabolites was increased predominantly in astrocytes whereas glutamatergic neurons were not affected. Previous results obtained using the same model [Neurosci. Lett. 279 (2000) 169] showed an increased turnover in both glutamatergic and GABAergic neurons 2 weeks after kainic acid injection. Combining the results from the two studies, it can be suggested that increased astrocytic activity 1 day after epileptic seizures results, subsequently, in an increased amino acid turnover in neurons.

Pyruvate recycling was studied in primary cultures of mouse cerebrocortical astrocytes, GABAergic cerebrocortical interneurons, and co-cultures consisting of both cell types by measuring production of [4-(13)C]glutamate from [3-(13)C]glutamate by aid of nuclear magnetic resonance spectroscopy. This change in the position of the label can only occur by entry of [3-(13)C]glutamate into the tricarboxylic acid (TCA) cycle, conversion of labeled alpha-ketoglutarate to malate or oxaloacetate, malic enzyme-mediated decarboxylation of malate to pyruvate or phosphoenolpyruvate carboxykinase-mediated conversion of oxaloacetate to phosphoenolpyruvate and subsequent hydrolysis of the latter to pyruvate, and introduction of the labeled pyruvate into the TCA cycle, i.e., after exit of the carbon skeleton of pyruvate from the TCA cycle followed by re-entry of the same pyruvate molecules via acetyl CoA. In agreement with earlier observations, pyruvate recycling was demonstrated in astrocytes, indicating the ability of these cells to undertake complete oxidative degradation of glutamate. The recycled [4-(13)C]glutamate was not further converted to glutamine, showing compartmentation of astrocytic metabolism. Thus, absence of recycling into glutamine in the brain in vivo cannot be taken as indication that pyruvate recycling is absent in astrocytes. No recycling could be demonstrated in the cerebrocortical neurons. This is consistent with a previously demonstrated lack of incorporation of label from glutamate into lactate, and it also indicates that mitochondrial malic enzyme is not operational. Nor was there any indication of pyruvate recycling in the co-cultures. Although this may partly be due to more rapid depletion of glutamate in the co-cultures, this observation at the very least indicates that pyruvate recycling is not up-regulated in the neuronal-astrocytic co-cultures.

Objective: To observe the effect of rat serum containing Buyanghuanwu decoction (BYHWT) on the proliferation of cultured rat cortical neurons, so as to understand the mechanism of BYHWT in the treatment of hypoxia brain damage., Methods: The growth of cultured rat cortical neurons were observed by MTT assay to evaluate the effect of the serum containing BYHWT on the neurons cultured in both normal and hypoxia conditions., Results: BYHWT significantly promoted proliferation of the neurons cultured under both normal and hypoxia conditions, in comparison with the response of the cells to drug-free serum (P<0.05)., Conclusion: Some of the constituents of BYHWT in rat serum can promote the proliferation of rat cortical neurons cultured in both normal and hypoxia conditions.

The interaction between astrocytes and neurons is examined from the standpoint of glutamate neurotoxicity. The review details 1. the distribution of glutamate transporters on astrocytes and neurons, provoking a reformulation of the interdependence between these two cell types in removing extracellular glutamate and preventing excitotoxic injury; 2. the potential involvement of aberrant glutamate transporter function in the etiology of neuropathological conditions; 3. the role of astrocyte-neuron interaction in widely divergent aspects of brain energetics; 4. the role of astrocytes in the process of glutamate recycling within the context of anesthetic treatment with pentobarbital and thiopental.

This study was performed to analyze the effects of glutamate and the epileptogenic agent pentylenetetrazole (PTZ) on neuronal glucose metabolism. Cerebellar granule neurons were incubated for 2 h in medium containing 3 mM [U-(13)C]glucose, with and without 0.25 mM glutamate and/or 10 mM PTZ. In the presence of PTZ, decreased glucose consumption with unchanged lactate release was observed, indicating decreased glucose oxidation. PTZ also slowed down tricarboxylic acid (TCA) cycle activity as evidenced by the decreased amounts of labeled aspartate and [1,2-(13)C]glutamate. When glutamate was present, glucose consumption was also decreased. However, the amount of glutamate, derived from [U-(13)C]glucose via the first turn of the TCA cycle, was increased. The decreased amount of [1,2-(13)C]glutamate, derived from the second turn in the TCA cycle, and increased amount of aspartate indicated the dilution of label due to the entrance of unlabeled glutamate into TCA cycle. In the presence of glutamate plus PTZ, the effect of PTZ was enhanced by glutamate. Labeled alanine was detected only in the presence of glutamate plus PTZ, which indicated that oxaloacetate was a better amino acid acceptor than pyruvate. Furthermore, there was also evidence for intracellular compartmentation of oxaloacetate metabolism. Glutamate and PTZ caused similar metabolic changes, however, via different mechanisms. Glutamate substituted for glucose as energy substrate in the TCA cycle, whereas, PTZ appeared to decrease mitochondrial activity.

Cerebellar granule neurons were incubated with or without glucose (3 mM) in the presence or absence of citrate (20 mM) using normoxic and/or hypoxic incubation conditions. During 4 h of hypoglycemia and also during hypoxia plus hypoglycemia, citrate increased lactate dehydrogenase (LDH) leakage from the cells and decreased mitochondrial activity, the latter was also the case in the presence of glucose. After 24 h of hypoglycemia, however, citrate decreased LDH leakage slightly, possibly due to its metabolism in the tricarboxylic acid cycle under these conditions. It should be noted that during mild hypoxia plus hypoglycemia a reduced LDH leakage was observed when compared to hypoglycemia alone. The 4 h low oxygen period did protect the neurons also during the 20 h re-oxygenation period. The present study might indicate that incubation of brain cell cultures in an atmosphere of air (30% oxygen) and 5% CO2, which is used in most laboratories, can be toxic and that oxygen concentration should be lowered considerably to mimic conditions in the brain.

Objective: To investigate the protective effect of Buyanghuanwu Tang (BYHWT), a decoction with 6 herbal ingredients, on in vitro cultured rat cortical neurons against apoptosis induced by hypoxia, and study its effect on the production of nitric oxide (NO) and oxygen free radicals and Bcl-2 expression in the neurons during hypoxia., Methods: Models of hypoxia-induced neuron apoptosis were established and treated with sera containing BYHWT. Stained by propidium iodide, the neurons underwent apoptosis analysis using flow cytometry, and the levels of malonyl dialdehyde (MDA) and NO were measured with spectrophotometer, with Bcl-2 expression assayed by immunohistochemical method with flow cytometry., Results: In BYHWT-treated neurons, apoptosis rates were significantly lower than those of the neurons subjected to hypoxia exclusively, and at the same time NO and MDA production was remarkably reduced. Bcl-2 expression, however, was up-regulated., Conclusion: BYHWT can protect neurons from hypoxia-induced apoptosis, the mechanism of which may lie in the elimination of NO and oxygen free radicals produced during hypoxia and up-regulation of Bcl-2 expression.

The aggregate-label learning paradigm tackles the long-standing temporary credit assignment (TCA) problem in neuroscience and machine learning, enabling spiking neural networks to learn multimodal sensory clues with delayed feedback signals. However, the existing aggregate-label learning algorithms only work for single spiking neurons, and with low learning efficiency, which limit their real-world applicability. To address these limitations, in this article, we first propose an efficient threshold-driven plasticity algorithm for spiking neurons, namely ETDP. It enables spiking neurons to generate the desired number of spikes that match the magnitude of delayed feedback signals and to learn useful multimodal sensory clues embedded within spontaneous spiking activities. Furthermore, we extend the ETDP algorithm to support multi-layer spiking neural networks (SNNs), which significantly improves the applicability of aggregate-label learning algorithms. We also validate the multi-layer ETDP learning algorithm in a multimodal computation framework for audio-visual pattern recognition. Experimental results on both synthetic and realistic datasets show significant improvements in the learning efficiency and model capacity over the existing aggregate-label learning algorithms. It, therefore, provides many opportunities for solving real-world multimodal pattern recognition tasks with spiking neural networks. [ABSTRACT FROM AUTHOR]

With a similar capability of processing spikes as biological neural systems, networks of spiking neurons are expected to achieve a performance similar to that of living brains. Despite the achievement of spiking neuron based applications, most of them assume noise-free condition for learning and testing. This assumption, though fairly general, ignores the fact that noise widely exists in spiking neural networks (SNNs) and the neural response can be significantly disturbed by noise. Therefore, how to deal with noise is an important issue in the applications of SNNs. Here, by analyzing strategies employed to make spiking neurons robust to noise, also inspired by biological neurons, we propose a strategy that train spiking neurons with a dynamic firing threshold named noise-threshold. The noise-threshold can be applied by the existing supervised learning methods to improve the noise tolerance of them. Experimental results show that, with a combination of noise-threshold, the anti-noise capability of the existing supervised learning methods improves significantly, and the trained neuron can precisely and reliably reproduce target sequences of spikes even under highly noisy conditions. More importantly, the SNNs-based computational model equipped with a noise-threshold is more robust and can achieve a good performance even with different types of noise. Therefore, the noise-threshold is significant to practical applications and theoretical researches of SNNs. [ABSTRACT FROM AUTHOR]

The spiking neural networks (SNNs) are the third generation of neural networks and perform remarkably well in cognitive tasks such as pattern recognition. The spike emitting and information processing mechanisms found in biological cognitive systems motivate the application of the hierarchical structure and temporal encoding mechanism in spiking neural networks, which have exhibited strong computational capability. However, the hierarchical structure and temporal encoding approach require neurons to process information serially in space and time respectively, which reduce the training efficiency significantly. For training the hierarchical SNNs, most existing methods are based on the traditional back-propagation algorithm, inheriting its drawbacks of the gradient diffusion and the sensitivity on parameters. To keep the powerful computation capability of the hierarchical structure and temporal encoding mechanism, but to overcome the low efficiency of the existing algorithms, a new training algorithm, the Normalized Spiking Error Back Propagation (NSEBP) is proposed in this paper. In the feedforward calculation, the output spike times are calculated by solving the quadratic function in the spike response model instead of detecting postsynaptic voltage states at all time points in traditional algorithms. Besides, in the feedback weight modification, the computational error is propagated to previous layers by the presynaptic spike jitter instead of the gradient decent rule, which realizes the layer-wised training. Furthermore, our algorithm investigates the mathematical relation between the weight variation and voltage error change, which makes the normalization in the weight modification applicable. Adopting these strategies, our algorithm outperforms the traditional SNN multi-layer algorithms in terms of learning efficiency and parameter sensitivity, that are also demonstrated by the comprehensive experimental results in this paper. [ABSTRACT FROM AUTHOR]

Abstract: Mesial temporal lobe epilepsy is associated with structural and functional abnormalities, such as hippocampal sclerosis and axonal reorganization. The temporal evolution of these changes remains to be determined, and there is a need for in vivo imaging techniques that can uncover the epileptogenic processes at an early stage. Manganese-enhanced magnetic resonance imaging may be useful in this regard. The aim of this study was to analyze the temporospatial changes in manganese enhancement in rat brain during the development of epilepsy subsequent to systemic kainate application (10 mg/kg i.p.). MnCl2 was given systemically on day 2 (early), day 15 (latent), and 11 weeks (chronic phase) after the initial status epilepticus. Twenty-four hours after MnCl2 injection T1-weighted 3D MRI was performed followed by analysis of manganese enhancement. In the medial temporal lobes, there was a pronounced decrease in manganese enhancement in CA1, CA3, dentate gyrus, entorhinal cortex and lateral amygdala in the early phase. In the latent and chronic phases, recovery of the manganese enhancement was observed in all these structures except CA1. A significant increase in manganese enhancement was detected in the entorhinal cortex and the amygdala in the chronic phase. In the latter phase, the structurally intact cerebellum showed significantly decreased manganese enhancement. The highly differentiated changes in manganese enhancement are likely to represent the net outcome of a number of pathological and pathophysiological events, including cell loss and changes in neuronal activity. Our findings are not consistent with the idea that manganese enhancement primarily reflects changes in glial cells. [Copyright &y& Elsevier]

Our understanding of the brain has developed from the theory that it is one continuous cell to the knowledge that there are many brain cells originally termed neurons and, furthermore to the discovery of glial cells and their multiple functions. Thus, an increasing complexity was unraveled and we have not reached a complete understanding of the phenomenon which comprises the compartmentation of metabolic pathways and metabolites. This is an important principle needed to fully understand the metabolic processes of the brain. At the cellular level this concept is well established whereas intracellular compartmentation has yet to be explored. Using magnetic resonance spectroscopy (MRS) for analysis of isotopomer composition combined with quantification of amino acid contents it is possible to construct models that describe intracellular compartmentation. Results of studies of cultures of astrocytes and neurons incubated in media containing [U-13C]glutamate in the presence or absence of thiopental may be used to propose an intracellular three compartment model of mitochondrial function. Due to the experimental paradigm only certain aspects of metabolism can be described. The present model consists of compartments assigned to CO2 production, glutamate synthesis from ketoglutarate and finally synthesis of a four-carbon metabolite which is shuttled between compartments. It is likely that metabolism may be far more complex than this and we are only beginning to glimpse some aspects of compartmentation at the cellular level. [Copyright &y& Elsevier]