Decoding grasp movement from monkey premotor cortex for real-time prosthetic hand control

Brain machine interfaces (BMIs) have demonstrated lots of successful arm-related reach decoding in past decades, which provide a new hope for restoring the lost motor functions for the disabled. On the other hand, the more sophisticated hand grasp movement, which is more fundamental and crucial for daily life, was less referred. Current state of arts has specified some grasp related brain areas and offline decoding results; however, online decoding grasp movement and real-time neuroprosthetic control have not been systematically investigated. In this study, we obtained neural data from the dorsal premotor cortex (PMd) when monkey reaching and grasping one of four differently shaped objects following visual cues. The four grasp gesture types with an additional resting state were classified asynchronously using a fuzzy k-nearest neighbor model, and an artificial hand was controlled online using a shared control strategy. The results showed that most of the neurons in PMd are tuned by reach and grasp movement, us- ing which we get a high average offline decoding accuracy of 97.1%. In the online demonstration, the instantaneous status of monkey grasping could be extracted successfully to control the artificial hand, with an event-wise accuracy of 85.1%. Overall, our results inspect the neural firing along the time course of grasp and for the first time enables asynchronous neural control of a prosthetic hand, which underline a feasible hand neural prosthesis in BMIs.     

基于运动相关脑电特征的手运动方向识别

为了研究如何从无创运动相关脑电中提取运动信息作为上肢主动康复训练的控制命令,通过设计实验,使右手完成左、上、右3个方向的运动,同时采集脑电数据和右手运动信息.通过小波时频分析确认与右手运动相关的脑电频带,并提取其小波分解系数作为特征,采用支持向量机进行特征分类,根据方向识别准确率分析提取特征的有效性.结果表明,运动脑电delta和theta频段的小波系数特征可以有效区分右手不同方向的运动,方向识别准确率的均值接近65%,并且用准备阶段特征分类的结果普遍优于运动阶段特征,因此,在手运动之前诱发的脑电活动含有丰富的运动信息,可用于脑-机接口系统提取上肢主动康复训练的控制命令.     

Cortical ensemble activity increasingly predicts behaviour outcomes during learning of a motor task

When an animal learns to make movements in response to different stimuli, changes in activity in the motor cortex seem to accompany and underlie this learning. The precise nature of modifications in cortical motor areas during the initial stages of motor learning, however, is largely unknown. Here we address this issue by chronically recording from neuronal ensembles located in the rat motor cortex, throughout the period required for rats to learn a reaction-time task. Motor learning was demonstrated by a decrease in the variance of the rats' reaction times and an increase in the time the animals were able to wait for a trigger stimulus. These behavioural changes were correlated with a significant increase in our ability to predict the correct or incorrect outcome of single trials based on three measures of neuronal ensemble activity: average firing rate, temporal patterns of firing, and correlated firing. This increase in prediction indicates that an association between sensory cues and movement emerged in the motor cortex as the task was learned. Such modifications in cortical ensemble activity may be critical for the initial learning of motor tasks.     

基于GA-SVM算法永磁同步直线电机变载荷进给位置误差预测

针对永磁同步直线电机精密进给过程中,受到齿槽效应、端部效应及摩擦力扰动等非线性因素的影响,位置误差难以预测问题,提出了一种基于遗传算法(GA)优化支持向量机(SVM)算法永磁直线电机变载荷位置误差预测模型的方法。通过测量各种情况下永磁直线电机在运动过程中的位置变化情况,利用遗传算法优化支持向量机算法建立预测模型。该模型采用实验台运行的正弦轨迹数据为训练样本,三角波轨迹数据为测试样本。选取各种情形的正弦波轨迹数据和三角波轨迹数据进行仿真预测和验证。以各种情况的正弦波信号的指令位置、指令速度和电流作为模型的输入,以三角波信号的位置误差作为输出。结果表明,经过遗传算法优化支持向量机建立的位置误差预测模型,在拟合和预测精度上要优于未经过算法优化的位置误差预测模型。     

鸽子运动意图的最大似然估计解码算法

神经信息解码是目前植入式脑机接口(brain-computer interface,BCI)神经信息处理研究中的难点和重点;解码效果的优劣以及解码算法的效率直接决定了脑机接口应用的有效性和实用性。为了解码十字迷宫内鸽子运动转向信息,利用高斯分布模型对神经元锋电位发放率的概率密度函数进行建模;并结合最大似然估计(maximum likelihood estimation,MLE)算法对鸽子的转向意图进行了预测;并将其结果与(support vector machine,SVM)法和群矢量(population vector,PV)法进行了比较。结果表明,MLE算法能够有效地解码鸽子的运动意图,解码正确率显著高于SVM法和PV法。这一结果也为进一步分析鸽子运动意图神经信息处理机制奠定了基础。     

基于深度学习的反无人机预警神经网络

为解决传统雷达探测设备面对"低小慢"无人机时产生的难检测与易突防问题,通过深度卷积神经网络对空中无人机进行实时识别,提取目标的类别与像空间位置信息;根据无人机像空间位置在时域下的变化趋势,绘制无人机飞行映射轨迹;利用长短期记忆网络对飞行映射轨迹进行预测,获取无人机在未来时域内的预测航迹方向,实现对无人机的预警跟踪、实时检测与轨迹推断。结果表明,所提出的算法中目标识别平均准确率可达到82%,轨迹预测平均准确率可达到80%计算速度可达到24帧/秒,可见能够在地基计算平台下对空中无人机进行实时精确预警,可以有效地防止识别领空内的非合作无人机渗透与突防。     

Relation of neurons in the nonprimary motor cortex to bilateral hand movement

In the primate cerebral cortex there are at least two somatotopically organized, nonprimary motor fields rostral to the primary motor area. To understand the functions of these multiple motor representations we have compared the neuronal activity in each of these fields while monkeys performed a trained motor task, using right, left or both hands. In the nonprimary motor cortex, activity in a number of neurons was related to the movement the animal chose and performed, whereas in the primary motor cortex, changes in the firing of most neurons were simply related to activity in the contralateral muscles. This result indicates that the nonprimary motor cortex is involved in higher-order coding of the laterality of the motor response, implying that it exerts its motor control function at a higher hierarchical level than its counterpart in the primary motor cortex.     

A neural correlate of response bias in monkey caudate nucleus

Primates are equipped with neural circuits in the prefrontal cortex, the parietal cortex and the basal ganglia that predict the availability of reward during the performance of behavioural tasks. It is not known, however, how reward value is incorporated in the control of action. Here we identify neurons in the monkey caudate nucleus that create a spatially selective response bias depending on the expected gain. In behavioural tasks, the monkey had to make a visually guided eye movement in every trial, but was rewarded for a correct response in only half of the trials. Reward availability was predictable on the basis of the spatial position of the visual target. We found that caudate neurons change their discharge rate systematically, even before the appearance of the visual target, and usually fire more when the contralateral position is associated with reward. Strong anticipatory activity of neurons with a contralateral preference is associated with decreased latency for eye movements in the contralateral direction. We conclude that this neuronal mechanism creates an advance bias that favours a spatial response when it is associated with a high reward value.     

穿戴式手枪射击稳定装置动力学仿真研究

为减小手枪射击时射击轨迹与瞄准基线的偏差,以子弹运动的方向为X轴正方向,与X轴垂直向上的方向为Y轴正方向设计了一种基于上臂外骨骼方案的手枪射击稳定装置。该装置通过限制手掌在X和Y方向上的运动及手腕关节处的转动来实现减小抖动的目的。分析了人体关节的阻抗类型,建立了手掌在射击前和射击后以及穿戴与未穿戴稳定装置状态下的运动方程,通过动力学仿真软件ADAMS计算手掌在X和Y方向上的速度和位移。得到的仿真结果表明方案达到了预期效果,并且在Y方向对位移的作用效果显著。     

无信号环形交叉口机非冲突机器学习预测方法

为高效精确地预测无信号环形交叉口机动车与非机动车的交通冲突,提出了基于遗传算法优化的BP神经网络(GA-BP)和支持向量回归(SVR)的组合预测模型(SVR-GA-BP)。通过无人机采集混合交通流高清视频,利用视频识别软件Tracker提取机非交通冲突轨迹数据,以距离碰撞时间(Time to Collision, TTC)为判别指标,确定机非冲突严重程度。基于偏相关性分析确定交通量、平均速度、大车比例等为机非交通冲突的显著影响因素,选取均方根误差(RMSE)、平均绝对误差(MAE)等五种评价指标对SVR模型、BP神经网络、SVR-GA-BP模型的预测值进行精度分析。结果表明,组合模型在一般冲突预测中精度为97.1%,相比SVR和BP神经网络分别提高6.9%和2.5%,在严重冲突预测中精度为96.1%,相比SVR和BP神经网络分别提高7.3%和5.1%。可见SVR-GA-BP组合模型能够有效预测无信号环形交叉口的机非冲突且精度最高,可为同类型交叉口的安全评价提供借鉴。     

Attention modulates synchronized neuronal firing in primate somatosensory cortex

A potentially powerful information processing strategy in the brain is to take advantage of the temporal structure of neuronal spike trains. An increase in synchrony within the neural representation of an object or location increases the efficacy of that neural representation at the next synaptic stage in the brain; thus, increasing synchrony is a candidate for the neural correlate of attentional selection. We investigated the synchronous firing of pairs of neurons in the secondary somatosensory cortex (SII) of three monkeys trained to switch attention between a visual task and a tactile discrimination task. We found that most neuron pairs in SII cortex fired synchronously and, furthermore, that the degree of synchrony was affected by the monkey's attentional state. In the monkey performing the most difficult task, 35% of neuron pairs that fired synchronously changed their degree of synchrony when the monkey switched attention between the tactile and visual tasks. Synchrony increased in 80% and decreased in 20% of neuron pairs affected by attention.     

Dissociation between hand motion and population vectors from neural activity in motor cortex

The population vector hypothesis was introduced almost twenty years ago to illustrate that a population vector constructed from neural activity in primary motor cortex (MI) of non-human primates could predict the direction of hand movement during reaching. Alternative explanations for this population signal have been suggested but could not be tested experimentally owing to movement complexity in the standard reaching model. We re-examined this issue by recording the activity of neurons in contralateral MI of monkeys while they made reaching movements with their right arms oriented in the horizontal plane-where the mechanics of limb motion are measurable and anisotropic. Here we found systematic biases between the population vector and the direction of hand movement. These errors were attributed to a non-uniform distribution of preferred directions of neurons and the non-uniformity covaried with peak joint power at the shoulder and elbow. These observations contradict the population vector hypothesis and show that non-human primates are capable of generating reaching movements to spatial targets even though population vectors based on MI activity do not point in the direction of hand motion.     

基于灰色神经网络的机床热误差建模

结合灰色模型和神经网络对数据处理的优点,提出了并联和嵌入型2种结构的灰色神经网络机床热误差预测模型。前者是在灰色模型和神经网络分别对机床热误差进行预测的基础上,采用线性组合方式,按照目标预测精度调整模型的加权系数,从而得到最终组合预测结果;后者是在神经网络输入层前增加灰化层,在输出层后增加白化层,通过对神经网络拓扑结构的改进,达到弱化原始数据随机性、提高预测模型鲁棒性和容错能力的目标。通过与传统灰色模型和神经网络进行试验结果对比表明：上述2种结构的灰色神经网络模型均提高了预测精度,且具有对原始数据要求低、计算简便、鲁棒性强等优点,可用于复杂实际加工场合中的数控机床热误差实时补偿。     

支持向量机和人工神经网络在期权价格预测中的比较研究

期权定价已成为金融市场的重要组成部分之一。 由于市场是动态的,准确预测期权价格非常困难。 因此,设计和发 展了各种机器学习技术来预测期权价格未来趋势。 比较了支持向量机(ＳＶＭ)模型和人工神经网络(ＡＮＮ)模型在期权价格预 测中的有效性。 在测试和训练阶段,２ 种模型都使用公开可用的基准数据集 ＳＰＹ ｏｐｔｉｏｎ ｐｒｉｃｅ－２０１５ 进行测试。 ２ 种模型均采 用主成分分析(ＰＣＡ)转换后的数据,以达到更好的预测精度。 另一方面,为了避免过拟合问题,将整个数据集划分为训练集 (７０％)和测试集(３０％)２ 组。 将支持向量机模型与基于均方根误差(ＲＭＳＥ)的神经网络模型的结果进行了比较。 实验结果 表明:神经网络模型优于支持向量机模型,预测的期权价格与相应的实际期权价格吻合良好。     

比较BP神经网络和RBF神经网络在基金净值预测中的应用

基金市场的活跃程度直接影响基金净值的变动,市场内部的影响因素具有较强的非线性特征,神经网络模型强大的非线性处理功能能够更为精准地预测基金净值的走势.本文采用BP神经网络和RBF神经网络对华夏成长基金进行实证分析,比较2种方法的预测精度.实证结果表明:RBF神经网络的仿真结果与真实值匹配程度较好,具有更高的预测精度.     

中华大蟾蜍胚后原始大脑皮层神经元自发电活动的发育变化

应用微电极电生理技术对中华大蟾蜍(Bufo bufo gargarizans)胚后端脑原始大脑皮层神经元的自发放电活动进行在体胞外记录,探讨其胚后端脑原始大脑皮层神经元自发电活的电生理学特性的发育变化。结果表明原始大脑皮层神经元的自发放电有5种形式。胚后发育的早期为4种放电类型,以簇状放电和连续放电为主;胚后发育的后期为5种放电类型,以簇状和连续簇状放电为主。随着原始大脑皮层的发育,单个放电的动作电位时程缩短,连续单个放电和连续簇状放电频率降低,连续放电、簇状放电和连续簇状放电的持续时间延长。随着端脑原始大脑皮层的发育,神经元的兴奋性逐步提高,神经元电活动形式逐渐呈现多样化。     

基于多元线性逆滤波器的视顶盖神经元集群亮度信息解码研究

通过解码鸽子视顶盖(the optic tectum,OT)神经元集群亮度信息的方法,研究了OT区神经元集群编码亮度信息的神经机制。首先设计了具有瞬态闪变特性的亮度视觉刺激模式,采用微电极阵列记录了多通道的锋电位(spike)发放序列;然后提取了神经元集群的spike发放率特征,构造多元线性逆滤波器解码视觉刺激亮度;最后采用互相关和信息论的方法对解码结果进行了分析。结果表明:采用的多元线性逆滤波器有效解码了OT区神经元集群编码的亮度信息。通过对解码参数变化与解码精度的分析可知神经元集群数目13个、bin的宽度5 ms、刺激后15 ms、刺激后持续时间35 ms具有最高的解码精度。通过神经元集群中逆滤波器形态的分析发现,单神经元在亮度信息的编码过程中受周围神经元的动态调制。通过对解码精度与刺激闪变频率关系的分析,发现在刺激闪变频率19 Hz到53 Hz的范围内,重建质量较好,在刺激闪变频率为33.5 Hz时,重建质量达到最优。     

基于机器学习与视觉的四足机器人步态精确监测方法

为保证四足机器人在复杂地形中能准确行进,对其步态实时监测方法进行了研究,提出一种基于机器学习和双目视觉的步态精确监测方法。首先为降低视觉残影误差,从已知步态轨迹“动中取静”确定对足端的最佳观测点位。此外,为补偿视觉系统误差所致足端位姿测量误差,提出一种基于深度神经网络的足端位姿精确预测方法。最后仿真结果表明所设计神经网络有99.68%的概率能达到0.025 mm足端位置预测精度,可满足实时、高精度监测要求。此方法将机器学习的泛化能力与视觉系统复杂误差来源相结合,使视觉方法实现了高精度测量,为足式机器人步态实时精确监测提供了新思路,进而为其足端精确定位及步态周期性保持提供了有益的方法参考。     

静液传动履带车辆转向神经元自适应PID控制

为实现静液传动履带车辆快速稳定转向,且转向轨迹可控,基于双侧轮边液压驱动结构特点,提出了转向时外侧马达排量采用压力、发动机转速双参数控制,内侧采用神经元自适应PID控制以跟随外侧的转向控制策略. 在Matlab/Simulink中建立了包含基于S函数的神经元PID控制器和综合控制策略Stateflow模块的整车模型,对转向控制进行仿真分析,阶跃输入时,神经元PID比传统PID控制能有效抑制系统超调量,加快系统响应速度;不同转向工况仿真结果表明:神经元PID控制具有较好的目标跟随能力,提高了系统的实时性和鲁棒性,使得静液传动履带车辆具有良好的转向性能.      

基于迁移学习和支持向量机的白细胞分类

针对人工镜检分类白细胞准确率和效率低的问题,基于深度学习和机器学习算法,提出了一种基于迁移学习和支持向量机的白细胞分类方法.首先对迁移模型进行微调训练,其次用微调训练后的迁移模型进行特征提取,然后将特征输入至神经网络和支持向量机中进行训练,最后通过神经网络和支持向量机的组合分类器对白细胞进行分类.实验结果表明,白细胞分类准确率由最初微调训练的83.26％,随着迁移模型的优化提升为90.43％,最后通过组合分类器再次提升为93.52％,可以在临床实践中帮助医生提高诊断的准确率和效率.