http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
On-Road Succeeding Vehicle Detection using Characteristic Visual Features
Shyam Prasad Adhikari(샴아디카리),Hitek Cho(조휘택),Hyeon-Joong Yoo(유현중),Changju Yang(양창주),Hyongsuk Kim(김형석) 대한전기학회 2010 전기학회논문지 Vol.59 No.3
A method for the detection of on-road succeeding vehicles using visual characteristic features like horizontal edges, shadow, symmetry and intensity is proposed. The proposed method uses the prominent horizontal edges along with the shadow under the vehicle to generate an initial estimate of the vehicle-road surface contact. Fast symmetry detection, utilizing the edge pixels, is then performed to detect the presence of vertically symmetric object, possibly vehicle, in the region above the initially estimated vehicle-road surface contact. A window defined by the horizontal and the vertical line obtained from above along with local perspective information provides a narrow region for the final search of the vehicle. A bounding box around the vehicle is extracted from the horizontal edges, symmetry histogram and a proposed squared difference of intensity measure. Experiments have been performed on natural traffic scenes obtained from a camera mounted on the side view mirror of a host vehicle demonstrate good and reliable performance of the proposed method.
Memristance Drift Avoidance with Charge Bouncing for Memristor-based Nonvolatile Memories
Shyam Prasad Adhikari,김형석,공배선,Leon O. Chua 한국물리학회 2012 THE JOURNAL OF THE KOREAN PHYSICAL SOCIETY Vol.61 No.9
A charge bouncing solution to avoid the undesirable drift in the programmed memory during readout of memristor-based non-volatile memory is proposed. Memristor memory can be programmed by strong programming signals, and the programmed memristance can be readout by weak readout signals. Though readout signals are weak compared to programming signals, readout charge is accumulated over time and leads to an undesirable drift of the operating point. This causes error in the programmed memory. Memristance drift is an important problem for practical utilization of memristors as memory. The only way presented so far to avoid drift is by converting input signals to doublets but un-ideal doublet signals still cause drifting problem. In the proposed method, drift is avoided with a capacitor in such a way that all the charge injected to the memristor during readout is stored in a capacitor, and bounced back through the memristor after completing the readout. Experimental results showing an excellent recovery from the temporal memristance drift using singlet pulses rather than the conventionally used doublet pulses for memristive memory readout are also presented.
2D Visual Features based Vehicle Detection
Shyam Prasad Adhikari(샴 아디카리),Sangmin Sim(심상민),hyongsuk Kim(김형석) 제어로봇시스템학회 2009 제어로봇시스템학회 합동학술대회 논문집 Vol.2009 No.12
This paper presents a monocular vehicle detection system using a combination of visual features like shadow underneath the vehicle, horizontal edges, symmetry and intensity, and not so rigid perspective constrains. The initial estimate of the vehicle position is provided by the horizontal edges along with shadow, which is further utilized for fast symmetry detection using edge pixels only. A narrow region is thus obtained where a final search gives a bounding box around a likely vehicle. Preliminary simulation results revealed good performance of the vehicle detection system.
A Circuit-Based Learning Architecture for Multilayer Neural Networks With Memristor Bridge Synapses
Adhikari, Shyam Prasad,Hyongsuk Kim,Budhathoki, Ram Kaji,Changju Yang,Chua, Leon O. IEEE 2015 IEEE Transactions on Circuits and Systems I: Regul Vol.62 No.1
<P>Memristor-based circuit architecture for multilayer neural networks is proposed. It is a first of its kind demonstrating successful circuit-based learning for multilayer neural network built with memristors. Though back-propagation algorithm is a powerful learning scheme for multilayer neural networks, its hardware implementation is very difficult due to complexities of the neural synapses and the operations involved in the learning algorithm. In this paper, the circuit of a multilayer neural network is designed with memristor bridge synapses and the learning is realized with a simple learning algorithm called Random Weight Change (RWC). Though RWC algorithm requires more iterations than back-propagation algorithm, we show that a circuit-based learning using RWC is two orders faster than its software counterpart. The method to build a multilayer neural network using memristor bridge synapses and a circuit-based learning architecture of RWC algorithm is proposed. Comparison between software-based and memristor circuit-based learning are presented via simulations.</P>
Hybrid no-propagation learning for multilayer neural networks
Adhikari, Shyam Prasad,Yang, Changju,Slot, Krzysztof,Strzelecki, Michal,Kim, Hyongsuk Elsevier 2018 Neurocomputing Vol.321 No.-
<P><B>Abstract</B></P> <P>A hybrid learning algorithm suitable for hardware implementation of multi-layer neural networks is proposed. Though backpropagation is a powerful learning method for multilayer neural networks, its hardware implementation is difficult due to complexities of the neural synapses and the operations involved in error backpropagation. We propose a learning algorithm with performance comparable to but easier than backpropagation to be implemented in hardware for on-chip learning of multi-layer neural networks. In the proposed learning algorithm, a multilayer neural network is trained with a hybrid of gradient-based delta rule and a stochastic algorithm, called Random Weight Change. The parameters of the output layer are learned using the delta rule, whereas the inner layer parameters are learned using Random Weight Change, thereby the overall multilayer neural network is trained without the need for error backpropagation. Experimental results showing better performance of the proposed hybrid learning rule than either of its constituent learning algorithms, and comparable to that of backpropagation on the benchmark MNIST dataset are presented. Hardware architecture illustrating the ease of implementation of the proposed learning rule in analog hardware vis-a-vis the backpropagation algorithm is also presented.</P>
Building cellular neural network templates with a hardware friendly learning algorithm
Adhikari, Shyam Prasad,Kim, Hyongsuk,Yang, Changju,Chua, Leon O. Elsevier 2018 Neurocomputing Vol.312 No.-
<P><B>Abstract</B></P> <P>A general solution for the construction of Cellular Neural Network (CNN) weights (cloning template) with Random Weight Change (RWC) algorithm is proposed. A target image for each input image is prepared via a sketch or any other kind of image processing technique for learning of Cellular Neural Network templates. A vector of randomly generated small values is added to the original weights and tested upon the input-target image pair. As a result, if the learning error decreases, the weight is taken for learning in the next iteration and updated using the same vector of random values. Otherwise, a new random vector for updating the weights is regenerated. One of the strong benefits of the proposed weight learning method is the simplicity of its learning algorithm and hence a simpler hardware architecture. Moreover the proposed method provides a unified solution to the problem of learning CNN templates without having to modify the original CNN structure and is applicable for all types of CNNs and input images. Successful learning of templates for various image processing tasks using different CNN structures are also demonstrated in this paper.</P>
샴아디카리(Shyam Prasad Adhikari),유현중(Hyeon-joong Yoo),김형석(Hyongsuk Kim) 제어로봇시스템학회 2011 제어·로봇·시스템학회 논문지 Vol.17 No.1
This paper presents a real- time detection of on-road succeeding vehicles based on low level edge features and a boosted cascade of Haar-like features. At first, the candidate vehicle location in an image is found by low level horizontal edge and symmetry characteristic of vehicle. Then a boosted cascade of the Haar-like features is applied to the initial hypothesized vehicle location to extract the refined vehicle location. The initial hypothesis generation using simple edge features speeds up the whole detection process and the application of a trained cascade on the hypothesized location increases the accuracy of the detection process. Experimental results on real world road scenario with processing speed of up to 27 frames per second for 720x480 pixel images are presented.
Mobile Robot Localization using Ceiling Landmark Positions and Edge Pixel Movement Vectors
Hongxin Chen(진홍신),Shyam Prasad Adhikari(아디카리 써얌프),Sungwoo Kim(김성우),Hyongsuk Kim(김형석) 제어로봇시스템학회 2010 제어·로봇·시스템학회 논문지 Vol.16 No.4
A new indoor mobile robot localization method is presented. Robot recognizes well designed single color landmarks on the ceiling by vision system, as reference to compute its precise position. The proposed likelihood prediction based method enables the robot to estimate its position based only on the orientation of landmark.The use of single color landmarks helps to reduce the complexity of the landmark structure and makes it easily detectable. Edge based optical flow is further used to compensate for some landmark recognition error. This technique is applicable for navigation in an unlimited sized indoor space. Prediction scheme and localization algorithm are proposed, and edge based optical flow and data fusing are presented. Experimental results show that the proposed method provides accurate estimation of the robot position with a localization error within a range of 5 cm and directional error less than 4 degrees.