How Dendrites Affect Online Recognition Memory.

In order to record the stream of autobiographical information that defines our unique personal history, our brains must form durable memories from single brief exposures to the patterned stimuli that impinge on them continuously throughout life. However, little is known about the computational strategies or neural mechanisms that underlie the brain's ability to perform this type of "online" learning. Based on increasing evidence that dendrites act as both signaling and learning units in the brain, we developed an analytical model that relates online recognition memory capacity to roughly a dozen dendritic, network, pattern, and task-related parameters. We used the model to determine what dendrite size maximizes storage capacity under varying assumptions about pattern density and noise level. We show that over a several-fold range of both of these parameters, and over multiple orders-of-magnitude of memory size, capacity is maximized when dendrites contain a few hundred synapses-roughly the natural number found in memory-related areas of the brain. Thus, in comparison to entire neurons, dendrites increase storage capacity by providing a larger number of better-sized learning units. Our model provides the first normative theory that explains how dendrites increase the brain's capacity for online learning; predicts which combinations of parameter settings we should expect to find in the brain under normal operating conditions; leads to novel interpretations of an array of existing experimental results; and provides a tool for understanding which changes associated with neurological disorders, aging, or stress are most likely to produce memory deficits-knowledge that could eventually help in the design of improved clinical treatments for memory loss.

Resonant-scanning dual-color STED microscopy with ultrafast photon counting: A concise guide.

STED (stimulated emission depletion) is a popular super-resolution fluorescence microscopy technique. In this paper, we present a concise guide to building a resonant-scanning STED microscope with ultrafast photon-counting acquisition. The STED microscope has two channels, using a pulsed laser and a continuous-wave (CW) laser as the depletion laser source, respectively. The CW STED channel preforms time-gated detection to enhance optical resolution in this channel. We use a resonant mirror to attain high scanning speed and ultrafast photon counting acquisition to scan a large field of view, which help reduce photobleaching. We discuss some practical issues in building a STED microscope, including creating a hollow depletion beam profile, manipulating polarization, and monitoring optical aberration. We also demonstrate a STED image enhancement method using stationary wavelet expansion and image analysis methods to register objects and to quantify colocalization in STED microscopy.

Ultrafast photon counting applied to resonant scanning STED microscopy.

To take full advantage of fast resonant scanning in super-resolution stimulated emission depletion (STED) microscopy, we have developed an ultrafast photon counting system based on a multigiga sample per second analogue-to-digital conversion chip that delivers an unprecedented 450 MHz pixel clock (2.2 ns pixel dwell time in each scan). The system achieves a large field of view (∼50 × 50 µm) with fast scanning that reduces photobleaching, and advances the time-gated continuous wave STED technology to the usage of resonant scanning with hardware-based time-gating. The assembled system provides superb signal-to-noise ratio and highly linear quantification of light that result in superior image quality. Also, the system design allows great flexibility in processing photon signals to further improve the dynamic range. In conclusion, we have constructed a frontier photon counting image acquisition system with ultrafast readout rate, excellent counting linearity, and with the capacity of realizing resonant-scanning continuous wave STED microscopy with online time-gated detection.

Darwin3: a large-scale neuromorphic chip with a novel ISA and on-chip learning.

Spiking neural networks (SNNs) are gaining increasing attention for their biological plausibility and potential for improved computational efficiency. To match the high spatial-temporal dynamics in SNNs, neuromorphic chips are highly desired to execute SNNs in hardware-based neuron and synapse circuits directly. This paper presents a large-scale neuromorphic chip named Darwin3 with a novel instruction set architecture, which comprises 10 primary instructions and a few extended instructions. It supports flexible neuron model programming and local learning rule designs. The Darwin3 chip architecture is designed in a mesh of computing nodes with an innovative routing algorithm. We used a compression mechanism to represent synaptic connections, significantly reducing memory usage. The Darwin3 chip supports up to 2.35 million neurons, making it the largest of its kind on the neuron scale. The experimental results showed that the code density was improved by up to 28.3× in Darwin3, and that the neuron core fan-in and fan-out were improved by up to 4096× and 3072× by connection compression compared to the physical memory depth. Our Darwin3 chip also provided memory saving between 6.8× and 200.8× when mapping convolutional spiking neural networks onto the chip, demonstrating state-of-the-art performance in accuracy and latency compared to other neuromorphic chips.

Dendritic Computing: Branching Deeper into Machine Learning.

In this paper, we discuss the nonlinear computational power provided by dendrites in biological and artificial neurons. We start by briefly presenting biological evidence about the type of dendritic nonlinearities, respective plasticity rules and their effect on biological learning as assessed by computational models. Four major computational implications are identified as improved expressivity, more efficient use of resources, utilizing internal learning signals, and enabling continual learning. We then discuss examples of how dendritic computations have been used to solve real-world classification problems with performance reported on well known data sets used in machine learning. The works are categorized according to the three primary methods of plasticity used-structural plasticity, weight plasticity, or plasticity of synaptic delays. Finally, we show the recent trend of confluence between concepts of deep learning and dendritic computations and highlight some future research directions.

ECG signal classification with binarized convolutional neural network.

Arrhythmias are a group of common conditions associated with irregular heart rhythms. Some of these conditions, for instance, atrial fibrillation (AF), might develop into serious syndromes if not treated in time. Therefore, for high-risk patients, early detection of arrhythmias is crucial. In this study, we propose employing deep convolutional neural network (CNN)-based algorithms for real-time arrhythmia detection. We first build a full-precision deep convolutional network model. With our proposed construction, we are able to achieve state-of-the-art level performance on the PhysioNet/CinC AF Classification Challenge 2017 dataset with our full-precision model. It is desirable to employ models with low computing resource requirements. It has been shown that a binarized model requires much less computing power and memory space than a full-precision model. We proceed to verify the feasibility of binarization in our neural network model. Network binarization can cause significant model performance degradation. Therefore, we propose employing a full-precision model as the teacher to regularize the training of the binarized model through knowledge distillation. With our proposed approach, we observe that network binarization only causes a small performance loss (the F1 score decreases from 0.88 to 0.87 for the validation set). Given that binarized convolutional networks can achieve favorable model performance while dramatically reducing computing cost, they are ideal for deployment on long-term cardiac condition monitoring devices. (Source code is available at https://github.com/yangfansun/bnn-ecg).

MUSTv2: An Improved De Novo Detection Program for Recently Active Miniature Inverted Repeat Transposable Elements (MITEs).

Background Miniature inverted repeat transposable element (MITE) is a short transposable element, carrying no protein-coding regions. However, its high proliferation rate and sequence-specific insertion preference renders it as a good genetic tool for both natural evolution and experimental insertion mutagenesis. Recently active MITE copies are those with clear signals of Terminal Inverted Repeats (TIRs) and Direct Repeats (DRs), and are recently translocated into their current sites. Their proliferation ability renders them good candidates for the investigation of genomic evolution. Results This study optimizes the C++ code and running pipeline of the MITE Uncovering SysTem (MUST) by assuming no prior knowledge of MITEs required from the users, and the current version, MUSTv2, shows significantly increased detection accuracy for recently active MITEs, compared with similar programs. The running speed is also significantly increased compared with MUSTv1. We prepared a benchmark dataset, the simulated genome with 150 MITE copies for researchers who may be of interest. Conclusions MUSTv2 represents an accurate detection program of recently active MITE copies, which is complementary to the existing template-based MITE mapping programs. We believe that the release of MUSTv2 will greatly facilitate the genome annotation and structural analysis of the bioOMIC big data researchers.

Resonant Scanning with Large Field of View Reduces Photobleaching and Enhances Fluorescence Yield in STED Microscopy.

Photobleaching is a major limitation of superresolution Stimulated Depletion Emission (STED) microscopy. Fast scanning has long been considered an effective means to reduce photobleaching in fluorescence microscopy, but a careful quantitative study of this issue is missing. In this paper, we show that the photobleaching rate in STED microscopy can be slowed down and the fluorescence yield be enhanced by scanning with high speed, enabled by using large field of view in a custom-built resonant-scanning STED microscope. The effect of scanning speed on photobleaching and fluorescence yield is more remarkable at higher levels of depletion laser irradiance, and virtually disappears in conventional confocal microscopy. With ≥6 GW∙cm(-2) depletion irradiance, we were able to extend the fluorophore survival time of Atto 647N and Abberior STAR 635P by ~80% with 8-fold wider field of view. We confirm that STED Photobleaching is primarily caused by the depletion light acting upon the excited fluorophores. Experimental data agree with a theoretical model. Our results encourage further increasing the linear scanning speed for photobleaching reduction in STED microscopy.

Capacity-enhancing synaptic learning rules in a medial temporal lobe online learning model.

Medial temporal lobe structures are responsible for recording the continuous stream of autobiographical memories that define our unique personal history. Remarkably, these areas can construct durable memories from brief exposures to the constantly changing activity patterns arriving from antecedent cortical areas. Using a computer model of the hippocampal Schaffer collateral pathway that incorporates evidence for dendritic spikes in CA1 pyramidal neurons, we searched for biologically-plausible long-term potentiation (LTP) and homeostatic depression (HD) rules that maximize "online" learning capacity. We found memory utilization is most efficient when (1) very few synapses are modified to store each pattern, (2) LTP, the learning operation, is dendrite-specific and gated by distinct pre- and postsynaptic thresholds, (3) HD, the forgetting operation, co-occurs with LTP and targets least-recently potentiated synapses, and (4) both LTP and HD are all-or-none, leading de facto to binary-valued synaptic weights. In networks containing up to 40 million synapses, the learning scheme led to order-of-magnitude capacity increases compared to conventional plasticity rules.

[Studies on properties of depotentiation of long-term potentiation induced by low frequency stimulation in CA1 neurons of rat hippocampal slices].

AIM AND METHODS: The parameters of low frequency stimulation (LFS) were altered systematically (frequencies of 1, 3 or 5 Hz; number of pulses of pulses of 300 or 900; and time lag after high frequency stimulation (HFS) of 20 or 100 min) and examined their effects on depotentiation (DP) of long-term potentiation (LTP) of synaptic transmission in CA1 neurons in hippocampal slices of rat. RESULTS: LTP could be induced by HFS (two trains of 100 Hz, 100 pulses, separated by 30 s) and be reversed to produce DP by a train of LFS of 900 pulses at 3 Hz given 20 min after HFS. DP induced by LFS could be blocked by NMDA receptor antagonist AP5 (50 micromol/L). And significantly reduced effect was observed for LFS at 1 Hz or 5 Hz, with smaller numbers of pulses or a longer time lag from LFS to HFS. CONCLUSION: The above results indicate that DP induced in CA1 neurons of rat hippocampal slices is strongly dependent on the parameters of LFS, and the process may be mediated through the NMDA receptor.

Activation of mitochondrial ATP-sensitive potassium channels delays ischemia-induced cellular uncoupling in rat heart.

AIM: To test the hypothesis that cellular uncoupling induced by myocardial ischemia is mediated by activation of mitochondrial ATP-sensitive potassium channels (mitoKATP). METHODS: Rat hearts were perfused on a Langendorff apparatus and subjected to 40-min ischemia followed by 30-min reperfusion (I/R). Changes in cellular coupling were monitored by measuring whole-tissue resistance. RESULTS: (1) In hearts subjected to I/R, the onset of uncoupling started at (13.3+/-1.0) min of ischemia; (2) Ischemic preconditioning (IPC) delayed the onset of uncoupling until (22.7+/-1.3) min. Blocking mitoKATP channels with 5-hydroxydecanoate (5-HD) before the IPC abolished the uncoupling delay [(12.6+/-1.6) min]; (3) Calcium preconditioning (CPC) had the same effect as IPC. And this effect was reversed by blocking the mitoKATP channel again. In the CPC group the onset of uncoupling occurred after (20.6+/-1.3) min, and this was canceled by 5-HD [(13.6+/-0.8) min]; (4) In hearts pretreated with the specific mitoKATP channel opener diazoxide before sustained ischemia, the onset was delayed to (18.4+/-1.4) min; (5) 5-HD canceled the protective effects of diazoxide (12.6+/-1.0) min; and both the L-type Ca2+ channel inhibitor verapamil and the free radical scavenger N-(2-mercaptopropionyl)glycine, reduced the extended onset time induced by diazoxide [to (13.3+/-1.8) min and (13.4+/-2.1) min, respectively]. CONCLUSION: IPC and CPC delay the onset of cellular uncoupling induced by acute ischemia in rat heart, and the underlying mechanism involves activation of the mitoKATP channels.