Advances in artificial intelligence (AI) have had a major impact on natural language processing (NLP), even more so with the emergence of large-scale language models like ChatGPT. This paper aims to provide a critical...
详细信息
For decades, the brain’s visual pathway has inspired machine and deep learning models, yet these models oversimplify the brain’s complex visual processing. This is manifested in the significant superiority of the br...
详细信息
For decades, the brain’s visual pathway has inspired machine and deep learning models, yet these models oversimplify the brain’s complex visual processing. This is manifested in the significant superiority of the brain in comparison to the developed models in terms of the accuracy and the amount of data needed for training. Therefore, rather than using the brain as an inspiration, in this paper, we introduce Img2Neuro;a convolutional neural network model feature extractor that predicts the visual brain’s response to images by encoding neural activity. Img2Neuro is trained on natural scene images paired with single-neuron recordings from the visual cortex and thalamus of mice and monkeys. We explore the feasibility of using Img2Neuro as a feature extractor for object recognition, where the output of Img2Neuro in response to unseen images is used as input to classifiers with the task of recognizing the object in the image. We evaluated our approach on three benchmark datasets;namely, MNIST, Fashion-MNIST, and Cifar10. In our experiments, we examined the classification performance when Img2Neuro is used as a feature extractor compared to using the images as direct input to the classifier, using five different classifiers;namely, linear discriminant analysis, perceptron, logistic regression, ridge classifier, and a single-layer neural network. The results demonstrate superior performance when using Img2Neuro in most datasets and across all classifiers, reaching an enhancement in accuracy of 9% on the MNIST dataset, 2% on FashionMNIST, and 18% on Cifar10 in some cases compared to using raw images as an input in the classifiers. The performance enhancements suggest that brain-trained encoders can effectively capture image features for object recognition tasks. By leveraging neural response data, Img2Neuro demonstrates a promising avenue for bridging the gap between biological and artificial visual processing, ultimately leading to novel strategies for improving state-of-t
While the $$H_\infty $$ observer-based control has found widespread application in the literature for integer-order systems, researchers have shown less interest in addressing the same issue within the fractional-orde...
While the $$H_\infty $$ observer-based control has found widespread application in the literature for integer-order systems, researchers have shown less interest in addressing the same issue within the fractional-order framework. In this context, this study delves into the application of $$H_\infty $$ observer-based control for the Hadamard fractional-order system (HFOS) described by the Takagi–Sugeno fuzzy models (TSFM). Using Lyapunov approach and by employing a matrix decoupling technique, LMI based conditions ensuring the existence of an observer and controller, are proposed. To minimize the impact of disturbances on the controlled output, $$H_\infty $$ optimization technique is used. The validity of our approach is substantiated through an example, underscoring the robustness and reliability of our proposed findings.
This paper describes a fuzzy neuron chip which is the modification of an ordinary neuron model by fuzzy logic, The algebraic product of scalar input and connective weights in synapse is replaced with a fuzzy inner pro...
详细信息
This paper describes a fuzzy neuron chip which is the modification of an ordinary neuron model by fuzzy logic, The algebraic product of scalar input and connective weights in synapse is replaced with a fuzzy inner product. An excitatory connection is represented by a MIN (minimum) operation and an inhibitory connection by fuzzy logic complement followed by a MIN operation;While an ordinary neuron model is established only by leaning, the fuzzy neuron is designable and optimized by learning, The fuzzy neuron is implemented in silicon wafer by a standard BICMOS process. The chip is applied to a handwritten character recognition system and it exhibits very high-speed recognition (less than 500 ns).
The plane elasticity problem of an infinite plate containing an elliptical inclusion is considered and the solutions for a point force and/or a dislocation located inside the inclusion are derived. By using the comple...
The plane elasticity problem of an infinite plate containing an elliptical inclusion is considered and the solutions for a point force and/or a dislocation located inside the inclusion are derived. By using the complex potential approach of Muskhelishvili, the general solutions are obtained in a form of a certain function plus an infinite series. The numerical convergence of the solutions is found to be better than that of Warren's solutions for the same problem. The proposed solutions are also appropriate for the case of a point force and dislocation acting at a point just on the interface.
In this paper, the singular stress fields created by an antiplane deformation at an inclusion corner are studied. It is shown that these singular stress fields can be separated into two independent types: a symmetric ...
In this paper, the singular stress fields created by an antiplane deformation at an inclusion corner are studied. It is shown that these singular stress fields can be separated into two independent types: a symmetric type with the stress singularity of 1/r(1-lambda 1) and a skew-symmetric type with the stress singularity of 1/r(1-lambda 2). These two types of the singular stress field can not occur simultaneously at the corner. If G(2) < G(1), there exists only the singularity of the skew-symmetric type, and if G(2) > G(1), there exists only the singularity of the symmetric type. A general expression of stress fields in the vicinity of the corner is presented. In the expression the singular stress fields for the symmetric type and the skew-symmetric type are defined in terms of the constants K-III, lambda(1) and K-III, lambda(2), respectively. K-III, lambda(1) and K-III, lambda(2) have to be determined from the complete boundary conditions of the given problem. For the problem of an infinite plate containing a diamond inclusion and subjected to a uniform longitudinal shear stress at infinity, the values of K-III, lambda(1) and K-III, lambda(2) are obtained by body force method. In the body force method, the investigated stresses are simulated by the superposition of the fundamental stress fields due to point forces, In order to obtain accurate solutions, the basic density functions of the distributed point forces are used, so that the stress singularities at the corner tip can be simulated by the point forces.
Establishing reliable correspondences is crucial for all registration tasks, including 2D image registration, 3D point cloud registration, and 2D-3D image-to-point cloud registration. However, these tasks are often co...
详细信息
Establishing reliable correspondences is crucial for all registration tasks, including 2D image registration, 3D point cloud registration, and 2D-3D image-to-point cloud registration. However, these tasks are often complicated by challenges such as scale inconsistencies, symmetry, and large deformations, which can lead to ambiguous matches. Previous feature-based and correspondence-based methods typically rely on geometric or semantic features to generate or polish initial potential correspondences. Some methods typically leverage specific geometric priors, such as topological preservation, to devise diverse and innovative strategies tailored to a given enhancement goal, which cannot be exhaustively enumerated. Additionally, many previous approaches rely on a single-step prediction head, which can struggle with local minima in complex matching scenarios. To address these challenges, we introduce an innovative paradigm that leverages a diffusion model in matrix space for robust matching matrix estimation. Our model treats correspondence estimation as a denoising diffusion process in the matching matrix space, gradually refining the intermediate matching matrix to the optimal one. Specifically, we apply the diffusion model in the doubly stochastic matrix space for 3D-3D and 2D-3D registration tasks. In the 2D image registration task, we deploy the diffusion model in a matrix sub-space, where dual-softmax projection regularization is applied. For all three registration tasks, we provide adaptive matching matrix embedding implementations tailored to the specific characteristics of each task while maintaining a consistent"match-to-warp" encoding pattern. Furthermore, we adopt a lightweight design for the denoising module. In inference, once points or image features are extracted and fixed, this module performs multi-step denoising predictions through reverse sampling. Evaluations on both 2D and 3D registration tasks demonstrate the effectiveness of our approach. Copyrigh
Memristor-based crossbar architectures have proven highly effective for matrix vector multiplication (MVM) operations, making them a promising solution for accelerating the MVMs widely used in precoding algorithms for...
详细信息
This paper investigates several properties of one-way alternating multicounter machines which operate in real time, and shows that (1) for each k greater-than-or-equal-to 1, one-way alternating k-counter machines (1 a...
详细信息
This paper investigates several properties of one-way alternating multicounter machines which operate in real time, and shows that (1) for each k greater-than-or-equal-to 1, one-way alternating k-counter machines (1 acm(k)'s) which operate in real time are less powerful than 1acm(k + 1)'s which operate in real time, (2) for each k greater-than-or-equal-to 2, 1acm(k)'s which operate in real time are less powerful than 1acm(k)'s which operate in linear time, and (3) for each k greater-than-or-equal-to 1, the class of sets accepted by 1acm(k)'s which operate in real time is not closed under concatenation with regular sets, Kleene closure, reversal and length-preserving homomorphism.
暂无评论