Winner-take-all Research Articles

The Separation of Powers in Ghana a Mirage? Examining the Excess Power of the Executive Arms of Government of Ghana

This study explores the extent to which the executive arm of government in Ghana exercises excessive power, undermining the principle of separation of powers and its implications for democratic governance. The objective is to evaluate the discretionary powers of the President in appointing key officials and to understand its impact on the autonomy of the legislative and judicial branches, political patronage, institutional effectiveness, and public trust. The methodology involved a comprehensive review of Ghana’s Fourth Republic Constitution (1992) and relevant literature to assess the legal framework and practical manifestations of executive dominance. The study critically analyzed the appointment powers vested in the President, including the roles of the Council of State, Parliament, and other advisory bodies. The results indicate that the President’s extensive appointment powers, as enshrined in the Constitution, significantly compromise the independence of other branches of government. Key appointments, such as those of the judiciary, ministers, and heads of public institutions, are heavily influenced by the executive, leading to a politicized public service and weakened institutional checks and balances. The practice of Winner-Takes-All (WTA) politics exacerbates this issue, fostering political exclusion and undermining democratic inclusivity. The conclusion highlights the need for constitutional reforms to decentralize appointment powers and enhance the role of independent bodies in the selection process. Strengthening parliamentary oversight and promoting merit-based appointments are recommended to ensure transparency, accountability, and the rule of law. Addressing these challenges is crucial for restoring public confidence in Ghana’s democratic institutions and promoting good governance.

Estimation of Wind Turbine Blade Icing Volume Based on Binocular Vision

Icing on wind turbine blades in cold and humid weather has become a detrimental factor limiting their efficient operation, and traditional methods for detecting blade icing have various limitations. Therefore, this paper proposes a non-contact ice volume estimation method based on binocular vision and improved image processing algorithms. The method employs a stereo matching algorithm that combines dynamic windows, multi-feature fusion, and reordering, integrating gradient, color, and other information to generate matching costs. It utilizes a cross-based support region for cost aggregation and generates the final disparity map through a Winner-Take-All (WTA) strategy and multi-step optimization. Subsequently, combining image processing techniques and three-dimensional reconstruction methods, the geometric shape of the ice is modeled, and its volume is estimated using numerical integration methods. Experimental results on volume estimation show that for ice blocks with regular shapes, the errors between the measured and actual volumes are 5.28%, 8.35%, and 4.85%, respectively; for simulated icing on wind turbine blades, the errors are 5.06%, 6.45%, and 9.54%, respectively. The results indicate that the volume measurement errors under various conditions are all within 10%, meeting the experimental accuracy requirements for measuring the volume of ice accumulation on wind turbine blades. This method provides an accurate and efficient solution for detecting blade icing without the need to modify the blades, making it suitable for wind turbines already in operation. However, in practical applications, it may be necessary to consider the impact of illumination and environmental changes on visual measurements.

Unsupervised end-to-end training with a self-defined target

Abstract Designing algorithms for versatile AI hardware that can learn on the edge using both labeled and unlabeled data is challenging. Deep end-to-end training methods incorporating phases of self-supervised and supervised learning are accurate and adaptable to input data but self-supervised learning requires even more computational and memory resources than supervised learning, too high for current embedded hardware. Conversely, unsupervised layer-by-layer training, such as Hebbian learning, is more compatible with existing hardware but doesn't integrate well with supervised learning. 
To address this, we propose a method enabling networks or hardware designed for end-to-end supervised learning to also perform high-performance unsupervised learning by adding two simple elements to the output layer: Winner-Take-All (WTA) selectivity and homeostasis regularization. These mechanisms introduce a 'self-defined target' for unlabeled data, allowing purely unsupervised training for both fully-connected and convolutional layers using backpropagation or equilibrium propagation on datasets like MNIST (up to 99.2%), Fashion-MNIST (up to 90.3%), and SVHN (up to 81.5%).
We extend this method to semi-supervised learning, adjusting targets based on data type, achieving 96.6% accuracy with only 600 labeled MNIST samples in a multi-layer perceptron. Our results show that this approach can effectively enable networks and hardware initially dedicated to supervised learning to also perform unsupervised learning, adapting to varying availability of labeled data.

Моделювання процесу ранжування у нейромережному класифікаторі об’єктів

As part of expert systems for various purposes, one of the basic ones is the decision support subsystem, which, in turn, requires the need for a procedure for classifying ob-jects. This is especially evident in intelligent medical diagnostic systems, which widely use artificial intelligence methods and tools. In this context, an approach involving mod-ern neurotechnology methods has proven to be effective at a high level. This paper con-siders a variant of the structural organization of a neural network classifier of objects as an improved model of the Hamming neural network. The peculiarity of this variant of the classifier is the expansion of its functionality by forming the ranks of the classified ob-ject in all defined classes. In the case of medical diagnosis, this means ranking all possi-ble diagnoses of a disease, i.e. determining not only the most likely diagnosis, but also the closest in rank to it. In fact, this will allow us to clarify the diagnosis, and thus im-prove the results of medical diagnosis. Accordingly, we simulated the classification pro-cess with the ranking of results, which corresponds to the classification with the realiza-tion of competition between the neurons of the competitive layer using negative-reverse (lateral) connections. This approach is basic in the theory of neural networks for deter-mining the winning neuron according to the WTA (Winner Takes All). Simulation model-ing of the classification variant was performed using specific biomedical data (eight symptoms) for the diagnosis of appendicitis (four diagnoses). The results of modeling the processes of neural network classification of objects with the formation of appropri-ate ranks are presented in the form of a table. They confirmed the correctness of the functioning algorithm for the considered classification model.

Inhibition SNN: unveiling the efficacy of various lateral inhibition learning in image pattern recognition

This study presents an enhanced spiking neural network (SNN) with inhibition, referred to as Inhibition SNN, which advances the field of pattern recognition through efficient event-driven computation. We incorporate a winner-take-all (WTA) mechanism into leaky integrate-and-fire neurons. This, combined with spike timing dependent plasticity, simulates the biological learning process. We propose two network prototypes, namely Inhibition V1 and Inhibition V2. Inhibition V1 differentiates between an excitatory layer and an inhibitory layer, while Inhibition V2 uses self-connections in the excitatory layer to implement WTA dynamics, enhancing signal contrast without a separate inhibitory layer. Our experiments on the MNIST dataset show that both Inhibition V1 and Inhibition V2 can match the accuracy of current advanced networks, with Inhibition V2 performing better. An unsupervised learning network of two layers (784–100) achieves 86% accuracy on MNIST and 61% on Fashion-MNIST. The scalability and robustness of these networks in neural computation underscore their potential for practical applications. This research offers a fresh perspective on SNN design, highlighting the impact of inhibition on learning efficiency. With plans to adapt these networks to neuromorphic hardware, Inhibition SNN could pave the way for energy-efficient intelligent systems.

Spike-based computational primitives to reproduce neural dynamics in the parietal cortex during motor preparation

Abstract Biologically plausible spiking neural network models of sensory cortices can be instrumental in understanding and validating their principles of computation. Models based on Cortical Computational Primitives (CCPs), such as Hebbian plasticity and Winner-Take-All (WTA) networks, have already been successful in this approach. However, the specific nature and roles of CCPs in sensorimotor cortices during cognitive tasks are yet to be fully deciphered. The evolution of motor intention in the Posterior Parietal Cortex (PPC) before arm-reaching movements is a well-suited cognitive process to assess the effectiveness of different CCPs. To this end, we propose a biologically plausible model composed of heterogeneous spiking neurons which implements and combines multiple CCPs, such as multi-timescale learning and soft WTA modules. By training the model to replicate the dynamics of in-vivo recordings from non-human primates, we show how it is effective in generating meaningful representations from unbalanced input data, and in faithfully reproducing the transition from motor planning to action selection. Our findings elucidate the importance of distributing spike-based plasticity across multi-timescales, and provide an explanation for the role of different CCPs in models of frontoparietal cortical networks for performing multisensory integration to efficiently inform action execution.

Memristor-based circuit design of BiLSTM network

The bidirectional long short-term memory (BiLSTM) network involves significant amount of parameter computations. This paper proposes the memristor-based bidirectional long short-term memory (MBiLSTM) network, with its capability of in-memory computing and parallel computing, can accelerates the parameter computations speed. The MBiLSTM network circuit is composed of normalization circuit, two memristor-based long short-term memory (LSTM) circuits, memristor-based resnet circuit, memristor-based dense circuitand winner-take-all (WTA) circuit. The voltage signals are scaled to the setting range by normalization circuit, memristor-based LSTM circuit is responsible for extracting features from the dataset, memristor-based resnet circuit can enhance the overall performance of the network, memristor-based dense circuit ensures that the final outputs dimension matches the dimension of the target signals, WTA circuit outputs the maximum voltage of memristor-based dense circuit. The effectiveness of the MBiLSTM network is validated through gait recognition experiment and handwritten digit recognition experiment. The stability, robustness and potential errors in the manufacturing process of memristance are analyzed.

Comparative Study of Different Classification Methods and Winner Takes All Approach

One of the most popular methods in remote sensing for gathering and evaluating satellite data is the classification of images. Several categories exist for image classification techniques, including supervised and unsupervised classification, pixel-based, object-based, and rule-based approaches. Each type of technique has pros and cons of its own. Choosing the method that produces the best results is one of the issues with image classification. The "best" model for classifying images relies on the particular task and the dataset used. The ideal classification technique is a crucial component in increasing classification accuracy. The strengths and drawbacks of various models vary, so selecting one that is appropriate for the job is critical. The main objective of this research is to analyze and compare the results of each classifier used, including ISODATA, K-mean, Maximum likelihood, Minimum distance, Support vector machine, and Neural network then integrate these different types of classification using the winners-takes-all classification approach in order to try to improve the results. The classified images were assessed, and both the overall accuracy and kappa coefficient were calculated and gave 79.50%, 73.89%, 77.05%, and 84.98%, 86.53%, 87.18%, and 88.69% for ISODATA, K-means, Minimum distance (MD), Maximum likelihood (MXL), Support vector machine (SVM), Neural network (NNT), and winner takes all (WTA), respectively. From the results, the Winner takes all (WTA) presented a superior in terms of the overall accuracy and kappa coefficient. Doi: 10.28991/CEJ-2024-010-10-016 Full Text: PDF

Influence of Imperfections on the Operational Correctness of DNN-kWTA Model.

The dual neural network (DNN)-based k -winner-take-all (WTA) model is able to identify the k largest numbers from its m input numbers. When there are imperfections, such as non-ideal step function and Gaussian input noise, in the realization, the model may not output the correct result. This brief analyzes the influence of the imperfections on the operational correctness of the model. Due to the imperfections, it is not efficient to use the original DNN- k WTA dynamics for analyzing the influence. In this regard, this brief first derives an equivalent model to describe the dynamics of the model under the imperfections. From the equivalent model, we derive a sufficient condition for which the model outputs the correct result. Thus, we apply the sufficient condition to design an efficiently estimation method for the probability of the model outputting the correct result. Furthermore, for the inputs with uniform distribution, a closed form expression for the probability value is derived. Finally, we extend our analysis for handling non-Gaussian input noise. Simulation results are provided to validate our theoretical results.

Improvement of horizontal streak on disparity map thru parameter optimization for stereo vision algorithm

In this paper, an improved local based stereo vision disparity map (SVDM) algorithm is proposed. The proposed local based SVDM algorithm include four stages and they are matching cost computation, cost aggregation disparity optimization and disparity refinement. The matching cost computation started by combining pixel to pixel matching techniques, which are absolute difference (AD) and gradient matching (GM) in producing the initial disparity map. Next, the cost aggregation uses minimum spanning tree (MST) segmentation, which equipped with edge preserving properties and noise filtering. Then, disparity optimization uses local approach with winner-take-all (WTA) technique. At the final stage, disparity refinement uses bilateral filter (BF) with weighted median (WM), which can improve the disparity map through noise removing and edges preserving. Then, the research continues to optimize the proposed local based SVDM algorithm through parameters optimization in obtaining the final disparity map. Here, multiple parameters from the proposed SVDM algorithm are manipulated and they are constant values for GM and several constant parameters in BF. By selecting the optimum parameter values, the performance of the proposed SVDM algorithm increased, especially robustness towards the horizontal streaks.

Synaptic ring attractor: A unified framework for attractor dynamics and multiple cues integration

Effective cue integration is essential for an animal's survival. The ring attractor network has emerged as a powerful framework for understanding how animals seamlessly integrate various cues. This network not only elucidates neural dynamics within the brain, especially in spatial encoding systems like the heading direction (HD) system, but also sheds light on cue integration within decision-making processes. Yet, many significant phenomena across different fields lack clear explanations. For instance, in physiology, the integration mechanism of Drosophila's compass neuron when confronted with conflicting self-motion cues and external sensory cues with varying gain control settings is not well elucidated. Similarly, in ethology, the decision-making system shows Bayesian integration (BI) under minimal cue conflicts, but shifts to a winner-take-all (WTA) mode as conflicts surpass a certain threshold. To address these gaps, we introduce a ring attractor network with asymmetrical neural connections and synaptic dynamics in this paper. A thorough series of simulations has been conducted to assess its ability to track external cues and integrate conflicting cues. The results from these simulations demonstrate that the proposed model replicates observed neural dynamics and offers a framework for modeling biologically plausible cue integration behaviors. Furthermore, our findings yield several testable predictions that could inform future neuroethological research, providing insights into the role of ring attractor dynamics in the animal brain.

An attention mechanism network based on the winner-take-all

In the field of neurology, neurons compete for brain attention in winner-take-all (WTA) competitions. Inspired by this, we propose a new WTA-based attention network (called ANW), which can be extended to general neural networks. The ANW network simulates WTA behavior in biological neurons; that is, the winner neuron itself becomes excited and sends its own spike signal. Furthermore, the winner will output an inhibition signal, which will be input into the cell synapses of other neurons. The ANW network regards the update of neuron weights after backpropagation as a stimulus input to neurons. Combined with the WTA function, the neural network will continue to update until it reaches a stable state. In this way, the ANW network can find the most active neurons in the network and generate the winners among these neurons to improve the ability of the neural network to extract features. As the ANW is a small and universal network, it can be seamlessly inserted into any neural network architecture, with negligible overhead, and can be trained together with a basic CNN. We verified the neural network with ANW by using the ImageNet-100, CIFAR-100 and MS COCO detection datasets for the experiments. The experimental results show that, compared with those of the baseline network, the top-1 err gains of 0.63% and 3.85% can be achieved for ANW, with classical ResNet-50 as the backbone network, respectively, demonstrating that the method can improve the performance of the model in image classification and target detection with wide applicability. The proposed method provides an effective, and almost consumption-neutral solution, for performance improvement in the field of computer vision target detection.

Stereo matching algorithm using deep learning and edge-preserving filter for machine vision

Machine vision research began with a single-camera system, but these systems had various limitations from having just one point-of-view of the environment and no depth information, therefore stereo cameras were invented. This paper proposes a hybrid method of a stereo matching algorithm with the goal of generating an accurate disparity map critical for applications such as 3D surface reconstruction and robot navigation to name a few. Convolutional neural network (CNN) is utilised to generate the matching cost, which is then input into cost aggregation to increase accuracy with the help of a bilateral filter (BF). Winner-take-all (WTA) is used to generate the preliminary disparity map. An edge-preserving filter (EPF) is applied to that output based on a transform that defines an isometry between curves on the 2D image manifold in 5D and the real line to eliminate these artefacts. The transform warps the input signal adaptively to allow linear 1D filtering. Due to the filter's resistance to high contrast and brightness, it is effective in refining and removing noise from the output image. Based on experimental research employing a Middlebury standard validation benchmark, this approach gives high accuracy with an average non-occluded error of 6.71% comparable to other published methods.

Refining disparity maps using deep learning and edge-aware smoothing filter

Stereo matching algorithm is crucial for applications that rely on three-dimensional (3D) surface reconstruction, producing a disparity map that contains depth information by computing the disparity values between corresponding points from a stereo image pair. In order to yield desirable results, the proposed stereo matching algorithm must possess a high degree of resilience against radiometric variation and edge inconsistencies. In this article convolutional neural network (CNN) is employed in the first stage to generate the raw matching cost, which is subsequently filtered with a bilateral filter (BF) and applied with cross-based cost aggregation (CBCA) during the cost aggregation stage to enhance precision. Winner-take-all (WTA) strategy is implemented to normalise the disparity map values. Finally, the resulting output is subjected to an edge-aware smoothing filter (EASF) to reduce the noise. Due to its resistance to high contrast and brightness, the filter is found to be effective in refining and eliminating noise from the output image. Despite discontinuities like adiron's lost cup handle or artl's shattered rods, this approach, based on experimental research utilizing a Middlebury standard validation benchmark, yields a high level of accuracy, with an average non-occluded error of 6.79%, comparable to other published methods.

Design of Multi-Competitors Winner-Take-All Neural Networks Based on DNA Strand Displacement for Molecular Pattern Recognition

DNA strand displacement technology (DSDT) provides flexible and powerful technical support for DNA molecular computing. DNA-based neural networks with Winner-Take-All (WTA) strategy has a great potential for nonlinear calculation. However, so far it has been limited to achieving the simultaneous competition of two competitors. Optimizing the calculation model and reducing system response time to recognize complex and functional molecular patterns remains a huge challenge. Here a novel neural network with WTA strategy based on DSDT was constructed, which allowed three competitors to participate in the competition at the same time. Firstly, the feasibility of the three-competitor WTA neural network was proved by 9-bit pattern recognition. Then the three-competitors WTA neural network was further extended to larger scale pattern recognition, which successfully recognized 64-bit letters A, B, and C and 100-bit handwritten digits 0, 2, and 4, respectively. Simulations showed that when recognizing the same target patterns with same number bits, compared with two-competitors WTA neural network, the three-competitors WTA network only used down to two-thirds DNA strands, and the system response time was reduced by more than ten times. This paper demonstrated the efficient recognition ability of the three-competitor WTA neural network, which is expected to be used to identify more complex information.

A study of stereo matching algorithm on low texture and depth discontinuity regions

<p>This article studies the performance of the proposed stereo matchingalgorithm on complex regions. These regions are areas with very limitedinformation for the matching process which are low texture, and depthdiscontinuity regions. In this study, each algorithm uses different matchingcost computation (MCC) techniques, but for cost aggregation (CA), disparityoptimization (DO) and disparity refinement (DR), the technique remains thesame. The MCC areabsolute difference(AD), the combination ofabsolutedifference and gradient matching(AD+GM) andcensus transform(CT).Then, for CA, DO and DR, they areminimum spanning tree(MST),winnertake all(WTA) andbilateral filter(BF), respectively. The results are presentedand discussed in this article. Hence, thru this study the robust method can beestimated at the MCC stage.</p>

Winner-Take-All and Loser-Take-All Circuits: Architectures, Applications and Analytical Comparison

Different winner-take-all (WTA) and loser-take-all (LTA) circuits are studied, and their operations are analyzed in this review. The exclusive operation of the current conveyor, binary tree, and time-domain WTA/LTA architectures, as the most important architectures reported in the literature, are compared from the perspectives of power consumption, speed, and precision.

Prototype-Based Interpretation of the Functionality of Neurons in Winner-Take-All Neural Networks.

Prototype-based learning (PbL) using a winner-take-all (WTA) network based on minimum Euclidean distance (ED-WTA) is an intuitive approach to multiclass classification. By constructing meaningful class centers, PbL provides higher interpretability and generalization than hyperplane-based learning (HbL) methods based on maximum inner product (IP-WTA) and can efficiently detect and reject samples that do not belong to any classes. In this article, we first prove the equivalence of IP-WTA and ED-WTA from a representational power perspective. Then, we show that naively using this equivalence leads to unintuitive ED-WTA networks in which the centers have high distances to data that they represent. We propose ±ED-WTA that models each neuron with two prototypes: one positive prototype, representing samples modeled by that neuron, and a negative prototype, representing the samples erroneously won by that neuron during training. We propose a novel training algorithm for the ±ED-WTA network, which cleverly switches between updating the positive and negative prototypes and is essential to the emergence of interpretable prototypes. Unexpectedly, we observed that the negative prototype of each neuron is indistinguishably similar to the positive one. The rationale behind this observation is that the training data that are mistaken for a prototype are indeed similar to it. The main finding of this article is this interpretation of the functionality of neurons as computing the difference between the distances to a positive and a negative prototype, which is in agreement with the BCM theory. Our experiments show that the proposed ±ED-WTA method constructs highly interpretable prototypes that can be successfully used for explaining the functionality of deep neural networks (DNNs), and detecting outlier and adversarial examples.

Optimal cardinal contests

We study the design of crowdsourcing contests in settings where the outputs of the contestants are quantifiable, for example, a data science challenge. This setting is in contrast to those where the output is only qualitative and cannot be objectively quantified, for example, when the goal of the contest is to design a logo. The literature on crowdsourcing contests focuses largely on ordinal contests, where contestants' outputs are ranked by the designer and awards are based on relative ranks. Such contests are ideally suited for the latter setting, where output is qualitative. For our setting (quantitative output), it is possible to design cardinal contests, where awards could be based on the actual outputs and not on their ranking alone—thus, the family of cardinal contests includes the family of ordinal contests. We study the problem of designing an optimal cardinal contest. We use mechanism design theory to derive an optimal cardinal mechanism and provide a convenient implementation—a decreasing reward‐meter mechanism—of the optimal contest. We establish the practicality of our mechanism by showing that it is “Obviously Strategy‐Proof,” a recently introduced formal notion of simplicity in the literature. We also compare the optimal cardinal contest with the most popular ordinal contest—namely, the Winner‐Takes‐All (WTA) contest, along several metrics. In particular, the optimal cardinal mechanism delivers a superior expected best output, whereas the WTA contest yields a greater expected contestant welfare. Furthermore, under a sufficiently large budget, the contest designer's expected net‐benefit is higher under the optimal cardinal mechanism than that under the WTA contest, regardless of the number of contestants in the two mechanisms. Our numerical analysis suggests that, for the contest designer, the average improvement provided by the optimal cardinal mechanism over the WTA contest is about 23%. For a given number of contestants, the benefit of the optimal cardinal mechanism is especially appreciable for projects where the ratio of the designer's utility to agents' cost‐of‐effort falls within a wide practical range. For projects where this ratio is very high, the expected profit of the best WTA contest is reasonably close to that of the optimal cardinal mechanism.

Effects of stochastic coding on olfactory discrimination in flies and mice.

Sparse coding can improve discrimination of sensory stimuli by reducing overlap between their representations. Two factors, however, can offset sparse coding's benefits: similar sensory stimuli have significant overlap and responses vary across trials. To elucidate the effects of these 2 factors, we analyzed odor responses in the fly and mouse olfactory regions implicated in learning and discrimination-the mushroom body (MB) and the piriform cortex (PCx). We found that neuronal responses fall along a continuum from extremely reliable across trials to extremely variable or stochastic. Computationally, we show that the observed variability arises from noise within central circuits rather than sensory noise. We propose this coding scheme to be advantageous for coarse- and fine-odor discrimination. More reliable cells enable quick discrimination between dissimilar odors. For similar odors, however, these cells overlap and do not provide distinguishing information. By contrast, more unreliable cells are decorrelated for similar odors, providing distinguishing information, though these benefits only accrue with extended training with more trials. Overall, we have uncovered a conserved, stochastic coding scheme in vertebrates and invertebrates, and we identify a candidate mechanism, based on variability in a winner-take-all (WTA) inhibitory circuit, that improves discrimination with training.