公开(公告)号：US5544280A

公开(公告)日：1996-08-06

申请号：US73018

申请日：1993-06-07

申请人： Hua-Kuang Liu ,  Jacob Barhen ,  Nabil H. Farhat ,  Chwan-Hwa Wu

发明人： Hua-Kuang Liu ,  Jacob Barhen ,  Nabil H. Farhat ,  Chwan-Hwa Wu

IPC分类号： G06N3/067 ,  G06F15/46

CPC分类号： G06N3/0675

摘要： A unipolar terminal-attractor based neural associative memory (TABAM) system with adaptive threshold for perfect convergence is presented. By adaptively setting the threshold values for the dynamic iteration for the unipolar binary neuron states with terminal-attractors for the purpose of reducing the spurious states in a Hopfield neural network for associative memory and using the inner-product approach, perfect convergence and correct retrieval is achieved. Simulation is completed with a small number of stored states (M) and a small number of neurons (N) but a large M/N ratio. An experiment with optical exclusive-OR logic operation using LCTV SLMs shows the feasibility of optoelectronic implementation of the models. A complete inner-product TABAM is implemented using a PC for calculation of adaptive threshold values to achieve a unipolar TABAM (UIT) in the case where there is no crosstalk, and a crosstalk model (CRIT) in the case where crosstalk corrupts the desired state.

摘要翻译： 提出了一种基于单极端子吸引子的神经相关记忆（TABAM）系统，具有完美收敛的自适应阈值。 通过自适应地设置用于具有终端吸引子的单极二进制神经元状态的动态迭代的阈值，以减少用于关联存储器的Hopfield神经网络中的伪状态并且使用内积法，完美收敛和正确检索是 实现了 模拟完成了少量存储状态（M）和少量神经元（N），但是M / N比较大。 使用LCTV SLM进行光学异或逻辑运算的实验显示了光电子实现模型的可行性。 使用PC实现完整的内部产品TABAM，用于计算自适应阈值以在没有串扰的情况下实现单极性TABAM（UIT），并且在串扰破坏所需状态的情况下使用串扰模型（CRIT） 。

公开(公告)号：US20100042266A1

公开(公告)日：2010-02-18

申请号：US10770857

申请日：2004-02-03

申请人： Jacob Barhen ,  Yehuda Y. Braiman ,  Vladimir Protopopescu

发明人： Jacob Barhen ,  Yehuda Y. Braiman ,  Vladimir Protopopescu

IPC分类号： G05D13/00 ,  G05B13/02

CPC分类号： C10M171/00

摘要： Methods and apparatus are described for control of friction at the nanoscale. A method of controlling frictional dynamics of a plurality of particles using non-Lipschitzian control includes determining an attribute of the plurality of particles; calculating an attribute deviation by subtracting the attribute of the plurality of particles from a target attribute; calculating a non-Lipschitzian feedback control term by raising the attribute deviation to a fractionary power ξ=(2m+1)/(2n+1) where n=1, 2, 3 . . . and m=0, 1, 2, 3 . . . , with m strictly less than n and then multiplying by a control amplitude; and imposing the non-Lipschitzian feedback control term globally on each of the plurality of particles; imposing causes a subsequent magnitude of the attribute deviation to be reduced.

摘要翻译： 描述了用于控制纳米级摩擦的方法和装置。 使用非Lipschitzian控制来控制多个粒子的摩擦动力学的方法包括确定多个粒子的属性; 通过从目标属性中减去多个粒子的属性来计算属性偏差; 通过将属性偏差提高到一个小数次幂来计算非Lipschitzian反馈控制项，并且其中n = 1,2,3，...（2m + 1）/（2n + 1）。 。 。 m = 0,1,2,3。 。 。 ，m严格小于n，然后乘以控制幅度; 以及在所述多个粒子的每一个上全局地施加非Lipschitzian反馈控制项; 强制性导致属性偏差的随后大小减小。

公开(公告)号：US07693587B2

公开(公告)日：2010-04-06

申请号：US10770857

申请日：2004-02-03

申请人： Jacob Barhen ,  Yehuda Y. Braiman ,  Vladimir Protopopescu

发明人： Jacob Barhen ,  Yehuda Y. Braiman ,  Vladimir Protopopescu

IPC分类号： G05B13/02 ,  G05B19/08 ,  G05B11/32 ,  G01P15/00 ,  B23K9/00

CPC分类号： C10M171/00

摘要： Methods and apparatus are described for control of friction at the nanoscale. A method of controlling frictional dynamics of a plurality of particles using non-Lipschitzian control includes determining an attribute of the plurality of particles; calculating an attribute deviation by subtracting the attribute of the plurality of particles from a target attribute; calculating a non-Lipschitzian feedback control term by raising the attribute deviation to a fractionary power ξ=(2m+1)/(2n+1) where n=1, 2, 3 . . . and m=0, 1, 2, 3 . . . , with m strictly less than n and then multiplying by a control amplitude; and imposing the non-Lipschitzian feedback control term globally on each of the plurality of particles; imposing causes a subsequent magnitude of the attribute deviation to be reduced.

摘要翻译： 描述了用于控制纳米级摩擦的方法和装置。 使用非Lipschitzian控制来控制多个粒子的摩擦动力学的方法包括确定多个粒子的属性; 通过从目标属性中减去多个粒子的属性来计算属性偏差; 通过将属性偏差提高到一个小数次幂来计算非Lipschitzian反馈控制项，并且其中n = 1,2,3，...（2m + 1）/（2n + 1）。 。 。 m = 0,1,2,3。 。 。 ，m严格小于n，然后乘以控制幅度; 以及在所述多个粒子的每一个上全局地施加非Lipschitzian反馈控制项; 强制性导致属性偏差的随后大小减小。

公开(公告)号：US5680515A

公开(公告)日：1997-10-21

申请号：US534537

申请日：1995-09-27

申请人： Jacob Barhen ,  Nikzad Toomarian ,  Amir Fijany ,  Michail Zak

发明人： Jacob Barhen ,  Nikzad Toomarian ,  Amir Fijany ,  Michail Zak

IPC分类号： G06F17/16 ,  G06N3/063 ,  G06F15/00

CPC分类号： G06F17/16 ,  G06N3/063

摘要： The present invention enhances the bit resolution of a CCD/CID MVM processor by storing each bit of each matrix element as a separate CCD charge packet. The bits of each input vector are separately multiplied by each bit of each matrix element in massive parallelism and the resulting products are combined appropriately to synthesize the correct product. In another aspect of the invention, such arrays are employed in a pseudo-spectral method of the invention, in which partial differential equations are solved by expressing each derivative analytically as matrices, and the state function is updated at each computation cycle by multiplying it by the matrices. The matrices are treated as synaptic arrays of a neural network and the state function vector elements are treated as neurons. In a further aspect of the invention, moving target detection is performed by driving the soliton equation with a vector of detector outputs. The neural architecture consists of two synaptic arrays corresponding to the two differential terms of the soliton-equation and an adder connected to the output thereof and to the output of the detector array to drive the soliton equation.

摘要翻译： 本发明通过将每个矩阵元素的每个比特存储为单独的CCD电荷分组来增强CCD / CID MVM处理器的比特分辨率。 每个输入向量的位分别乘以每个矩阵元素的每个比特的大规模并行性，并将所得到的乘积合并以合成正确的乘积。 在本发明的另一方面，这种阵列用于本发明的伪光谱方法，其中通过将每个导数分析地表示为矩阵来求解偏微分方程，并且通过将每个计算周期的状态函数乘以 矩阵。 矩阵被视为神经网络的突触阵列，状态函数向量元素被视为神经元。 在本发明的另一方面，通过用检测器输出的向量驱动孤子方程来执行移动目标检测。 神经体系结构由两个突变阵列组成，这两个突触阵列对应于孤子方程的两个微分项，以及连接到其输出的加法器和检测器阵列的输出以驱动孤子方程。

公开(公告)号：US5508538A

公开(公告)日：1996-04-16

申请号：US161908

申请日：1993-11-30

申请人： Amir Fijany ,  Jacob Barhen ,  Nikzad Toomarian

发明人： Amir Fijany ,  Jacob Barhen ,  Nikzad Toomarian

IPC分类号： G06F17/16 ,  G06N3/063 ,  H01L29/76 ,  G06G7/00 ,  G11C19/18 ,  H01L27/148

CPC分类号： G06F17/16 ,  G06N3/063

摘要： The present invention is embodied in a charge coupled device (CCD)/charge injection device (CID) architecture capable of performing a Fourier transform by simultaneous matrix vector multiplication (MVM) operations in respective plural CCD/CID arrays in parallel in O(1) steps. For example, in one embodiment, a first CCD/CID array stores charge packets representing a first matrix operator based upon permutations of a Hartley transform and computes the Fourier transform of an incoming vector. A second CCD/CID array stores charge packets representing a second matrix operator based upon different permutations of a Hartley transform and computes the Fourier transform of an incoming vector. The incoming vector is applied to the inputs of the two CCD/CID arrays simultaneously, and the real and imaginary parts of the Fourier transform are produced simultaneously in the time required to perform a single MVM operation in a CCD/CID array.

公开(公告)号：US06188964B1

公开(公告)日：2001-02-13

申请号：US09395536

申请日：1999-09-14

申请人： David B. Reister ,  Jacob Barhen ,  Edward M. Oblow

发明人： David B. Reister ,  Jacob Barhen ,  Edward M. Oblow

IPC分类号： G01V128

CPC分类号： G01V1/362 ,  G01V2210/53

摘要： An efficient method for generating residual statics corrections to compensate for surface-consistent static time shifts in stacked seismic traces. The method includes a step of framing the residual static corrections as a global optimization problem in a parameter space. The method also includes decoupling the global optimization problem involving all seismic traces into several one-dimensional problems. The method further utilizes a Stochastic Pijavskij Tunneling search to eliminate regions in the parameter space where a global minimum is unlikely to exist so that the global minimum may be quickly discovered. The method finds the residual statics corrections by maximizing the total stack power. The stack power is a measure of seismic energy transferred from energy sources to receivers.

摘要翻译： 一种用于产生残余静力学校正以补偿堆叠地震轨迹中表面一致的静态时间偏移的有效方法。 该方法包括将剩余静态校正作为参数空间中的全局优化问题进行成帧的步骤。 该方法还包括将涉及所有地震轨迹的全局优化问题解耦为若干一维问题。 该方法还利用随机Pijavskij隧道搜索来消除参数空间中不可能存在全局最小值的区域，从而可以快速发现全局最小值。 该方法通过最大化总堆叠功率来找到残差静态校正。 堆叠功率是从能量传输到接收器的地震能量的量度。

公开(公告)号：US5952685A

公开(公告)日：1999-09-14

申请号：US598900

申请日：1996-02-09

申请人： Amir Fijany ,  Jacob Barhen ,  Nikzad Toomarian

发明人： Amir Fijany ,  Jacob Barhen ,  Nikzad Toomarian

IPC分类号： G06F17/16 ,  G06N3/063 ,  H01L29/76 ,  G06G7/00 ,  G11C19/18 ,  H01L27/148

CPC分类号： G06F17/16 ,  G06N3/063

摘要： The present invention is embodied in a charge coupled device (CCD)/charge injection device (CID) architecture capable of performing a Fourier transform by simultaneous matrix vector multiplication (MVM) operations in respective plural CCD/CID arrays in parallel in O(1) steps. For example, in one embodiment, a first CCD/CID array stores charge packets representing a first matrix operator based upon permutations of a Hartley transform and computes the Fourier transform of an incoming vector. A second CCD/CID array stores charge packets representing a second matrix operator based upon different permutations of a Hartley transform and computes the Fourier transform of an incoming vector. The incoming vector is applied to the inputs of the two CCD/CID arrays simultaneously, and the real and imaginary parts of the Fourier transform are produced simultaneously in the time required to perform a single MVM operation in a CCD/CID array.

摘要翻译： 本发明体现在能够在O（1）中并行地在多个CCD / CID阵列中通过同时进行矩阵矢量乘法（MVM）运算的傅里叶变换的电荷耦合器件（CCD）/电荷注入器件（CID） 脚步。 例如，在一个实施例中，第一CCD / CID阵列基于哈特利变换的置换来存储表示第一矩阵运算符的电荷分组，并计算输入向量的傅里叶变换。 基于哈特利变换的不同排列，第二CCD / CID阵列存储表示第二矩阵运算符的电荷分组，并计算进入矢量的傅里叶变换。 输入矢量同时应用于两个CCD / CID阵列的输入，傅立叶变换的实部和虚部在CCD / CID阵列中执行单个MVM操作所需的时间内同时产生。

公开(公告)号：US5930781A

公开(公告)日：1999-07-27

申请号：US969868

申请日：1992-10-27

申请人： Nikzad Toomarian ,  Jacob Barhen

发明人： Nikzad Toomarian ,  Jacob Barhen

IPC分类号： G06N3/04 ,  G06N3/08 ,  G06F15/18

CPC分类号： G06N3/049 ,  G06N3/08

摘要： A method and apparatus for supervised neural learning of time dependent trajectories exploits the concepts of adjoint operators to enable computation of the gradient of an objective functional with respect to the various parameters of the network architecture in a highly efficient manner. Specifically, it combines the advantage of dramatic reductions in computational complexity inherent in adjoint methods with the ability to solve two adjoint systems of equations together forward in time. Not only is a large amount of computation and storage saved, but the handling of real-time applications becomes also possible. The invention has been applied it to two examples of representative complexity which have recently been analyzed in the open literature and demonstrated that a circular trajectory can be learned in approximately 200 iterations compared to the 12000 reported in the literature. A figure eight trajectory was achieved in under 500 iterations compared to 20000 previously required. The trajectories computed using our new method are much closer to the target trajectories than was reported in previous studies.

摘要翻译： 用于时间依赖轨迹的监督神经学习的方法和装置利用伴随运算符的概念以使得能够以高效的方式相对于网络架构的各种参数来计算目标函数的梯度。 具体来说，它结合了伴随方法固有的计算复杂度的显着降低的优点，以及在时间上一起解决两个方程组合的能力。 不仅节省了大量的计算和存储空间，而且实时应用程序的处理也是可能的。 本发明已经应用于两个最近在开放文献中分析的代表性复杂性的例子，并且证明了与文献中报道的12000相比，可以在大约200次迭代中学习圆形轨迹。 在500次迭代之前实现了图8的轨迹，相比之前需要的20000次。 使用我们的新方法计算的轨迹比以前的研究报道的更接近目标轨迹。

公开(公告)号：US5491650A

公开(公告)日：1996-02-13

申请号：US49829

申请日：1993-04-19

申请人： Jacob Barhen ,  Nikzad Toomarian ,  Amir Fijany ,  Michail Zak

发明人： Jacob Barhen ,  Nikzad Toomarian ,  Amir Fijany ,  Michail Zak

IPC分类号： G06F17/16 ,  G06N3/063 ,  G06J1/00 ,  G06F7/52

CPC分类号： G06F17/16 ,  G06N3/063

摘要： The present invention discloses increased bit resolution of a charge coupled device (CCD)/charge injection device (CID) matrix vector multiplication (MVM) processor by storing each bit of each matrix element as a separate CCD charge packet. The bits of each input vector are separately multiplied by each bit of each matrix element in massive parallelism and the resulting products are combined appropriately to synthesize the correct product. In addition, such arrays are employed in a pseudo-spectral method of the invention, in which partial differential equations are solved by expressing each derivative analytically as matrices, and the state function is updated at each computation cycle by multiplying it by the matrices. The matrices are treated as synaptic arrays of a neutral network and the state function vector elements are treated as neurons. Further, moving target detection is performed by driving the soliton equation with a vector of detector outputs. The neural architecture consists of two synaptic arrays corresponding to the two differential terms of the soliton equation and an adder connected to the output thereof and to the output of the detector array to drive the soliton equation.

摘要翻译： 本发明通过将每个矩阵元素的每一位存储为单独的CCD电荷包，提高了电荷耦合器件（CCD）/电荷注入器件（CID）矩阵矢量乘法（MVM）处理器的位分辨率。 每个输入向量的位分别乘以每个矩阵元素的每个比特的大规模并行性，并将所得到的乘积合并以合成正确的乘积。 此外，在本发明的伪光谱方法中采用这样的阵列，其中通过将每个导数分析地表示为矩阵来求解偏微分方程，并且通过将其乘以矩阵在每个计算周期更新状态函数。 矩阵被视为中性网络的突触阵列，状态函数向量元素被视为神经元。 此外，通过用检测器输出的向量驱动孤子方程来执行移动目标检测。 神经结构由两个对应于孤子方程的两个微分项的突触阵列和连接到其输出的加法器和检测器阵列的输出驱动孤子方程组成。

公开(公告)号：US5428710A

公开(公告)日：1995-06-27

申请号：US908677

申请日：1992-06-29

申请人： Nikzad Toomarian ,  Jacob Barhen

发明人： Nikzad Toomarian ,  Jacob Barhen

IPC分类号： G06N3/08 ,  G06F15/18

CPC分类号： G06N3/08

摘要： A neural network is trained to output a time dependent target vector defined over a predetermined time interval in response to a time dependent input vector defined over the same time interval by applying corresponding elements of the error vector, or difference between the target vector and the actual neuron output vector, to the inputs of corresponding output neurons of the network corrective feedback. This feedback decreases the error and quickens the learning process, so that a much smaller number of training cycles are required to complete the learning process. A conventional gradient descent algorithm is employed to update the neural network parameters at the end of the predetermined time interval. The foregoing process is repeated in repetitive cycles until the actual output vector corresponds to the target vector. In the preferred embodiment, as the overall error of the neutral network output decreases during successive training cycles, the portion of the error fed back to the output neurons is decreased accordingly, allowing the network to learn with greater freedom from teacher forcing as the network parameters converge to their optimum values. The invention may also be used to train a neural network with stationary training and target vectors.

摘要翻译： 训练神经网络以响应于在相同时间间隔上定义的时间相关输入向量来输出在预定时间间隔上定义的时间相关目标矢量，通过应用误差向量的相应元素，或者目标矢量与实际 神经元输出向量，输入到相应输出神经元的网络校正反馈。 这种反馈减少了错误，加快了学习过程，从而完成学习过程所需的训练周期要少得多。 采用常规梯度下降算法在预定时间间隔结束时更新神经网络参数。 以重复循环重复上述过程，直到实际输出向量对应于目标矢量。 在优选实施例中，随着中继网络输出的总体误差在连续的训练周期中减小，反馈给输出神经元的误差部分相应地减小，从而允许网络以较大的自由度学习从教师强迫中获得的网络参数 收敛到其最佳值。 本发明还可以用于训练具有固定训练和目标矢量的神经网络。