178 results about "Synaptic weight" patented technology

A spiking neural network has a layer of connected neurons exchanging signals. Each neuron is connected to at least one other neuron. A neuron is active if it spikes at least once during a time interval. Time-varying synaptic weights are computed between each neuron and at least one other neuron connected thereto. These weights are computed according to a number of active neurons that are connected to the neuron. The weights are also computed according to an activity of the spiking neural network during the time interval. Spiking of each neuron is synchronized according to a number of active neurons connected to the neuron and according to the weights. A pattern is submitted to the spiking neural network for generating sequences of spikes, which are modulated over time by the spiking synchronization. The pattern is characterized according to the sequences of spikes generated in the spiking neural network.

The invention provides an operation device and method of an accelerating chip which accelerates a depth neural network algorithm. The device comprises a vector addition processor module which carries out addition or subtraction of vectors and / or the vector operation of a pooling layer algorithm in the depth neural network algorithm, a vector function value operator module which carries out the vector operation of the nonlinear evaluation in the depth neural network algorithm, and a vector multiply adder module which carries out the multiply addition operation of the vectors. The three modules execute a programmable instruction and interact with each other to calculate a neural network output result and a synaptic weight change amount which represents the neuron action intensity between neural layers. The three modules are provided with middle value storage areas and carry out reading and writing operations on a master memory. Thus, the reading and writing frequencies of the middle value of the master memory can be reduced, the energy consumption of the accelerator chip can be reduced, and the data loss and replacement problems in a data processing process can be avoided.

The invention discloses a convolution neural network on-chip learning system based on non-volatile memory, comprising an input module, a convolution neural network module, an output module and a weight update module. The on-chip learning of the convolution neural network module utilizes the characteristic that the conductance of the memristor changes with the applied pulse to realize the synapticfunction, and the convolution kernel value or the synaptic weight value is stored in the memristor unit. The input module converts the input signal into the voltage signal required by the convolutional neural network module. The convolutional neural network module transforms the input voltage signal layer by layer and transmits the result to the output module to get the output of the network. Theweight updating module adjusts the conductance value of the memristor in the convolutional neural network module according to the result of the output module, and updates the convolution core value orsynaptic weight value of the network. The invention aims at realizing the on-chip learning of the convolution neural network, realizing the on-line processing of the data, and realizing the requirements of high speed, low power consumption and low hardware cost based on the high parallelism of the network.

A method and apparatus for identifying running vehicles in an area to be monitored using acoustic signature recognition. The apparatus includes an input sensor for capturing an acoustic waveform produced by a vehicle source, and a processing system. The waveform is digitized and divided into frames. Each frame is filtered into a plurality of gammatone filtered signals. At least one spectral feature vector is computed for each frame. The vectors are integrated across a plurality of frames to create a spectro-temporal representation of the vehicle waveform. In a training mode, values from the spectro-temporal representation are used as inputs to a Nonlinear Hebbian learning function to extract acoustic signatures and synaptic weights. In an active mode, the synaptic weights and acoustic signatures are used as patterns in a supervised associative network to identify whether a vehicle is present in the area to be monitored. In response to a vehicle being present, the class of vehicle is identified. Results may be provided to a central computer.

A Wings array system for communicating between nodes using store and load instructions is described. Couplings between nodes are made according to a 1 to N adjacency of connections in each dimension of a G×H matrix of nodes, where G≧N and H≧N and N is a positive odd integer. Also, a 3D Wings neural network processor is described as a 3D G×H×K network of neurons, each neuron with an N×N×N array of synaptic weight values stored in coupled memory nodes, where G≧N, H≧N, K≧N, and N is determined from a 1 to N adjacency of connections used in the G×H×K network. Further, a hexagonal processor array is organized according to an INFORM coordinate system having axes at 60 degree spacing. Nodes communicate on row paths parallel to an FM dimension of communication, column paths parallel to an IO dimension of communication, and diagonal paths parallel to an NR dimension of communication.

Certain embodiments of the present disclosure support techniques for training of synapses in biologically inspired networks. Only one device based on a memristor can be used as a synaptic connection between a pair of neurons. The training of synaptic weights can be achieved with a low current consumption. A proposed synapse training circuit may be shared by a plurality of incoming / outgoing connections, while only one digitally implemented pulse-width modulation (PWM) generator can be utilized per neuron circuit for generating synapse-training pulses. Only up to three phases of a slow clock can be used for both the neuron-to-neuron communications and synapse training. Some special control signals can be also generated for setting up synapse training events. By means of these signals, the synapse training circuit can be in a high-impedance state outside the training events, thus the synaptic resistance (i.e., the synaptic weight) is not affected outside the training process.