Conversion represents an effective approach for obtaining low-power models by transforming Artificial Neural Networks (ANNs) into event-driven Spiking Neural Networks (SNNs) without additional training. However, existing training-free conversion methods often incur substantial conversion errors. Here, we first reveal that these conversion errors primarily arise from a distributional mismatch, as the activation distributions of ANNs exhibit channel-wise shifts and scaling, whereas spike rates lack corresponding channel-specific characteristics. To address this limitation, we propose Adaptive Integrate-and-Fire (AIF) neurons with channel-specific thresholds and membrane-potential offsets that dynamically adjust spike rates. These parameters are optimized to jointly minimize conversion errors and maximize information entropy, enabling AIF neurons to capture the activation distribution characteristics of the original ANN. Moreover, AIF neurons can be seamlessly integrated into Transformer architectures with only negligible additional computational cost. Our method achieves state-of-the-art results on multiple vision and natural language processing benchmarks, in particular attaining a notable top-1 accuracy of 85.52% on ImageNet-1K.

Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks, but their deployment in resource-constrained environments remains challenging due to substantial memory and computational requirements. Benefiting from the sparse event-driven computation paradigm of Spiking Neural Networks (SNNs), some research has focused on designing spike-based language models. However, existing spike-based language models achieve only partial computational efficiency gains and fail to address memory constraints comprehensively. In this paper, we propose an evolved and quantized spike-driven language model (EQ-SpikeLM) to address identified challenges. This model incorporates two primary innovations. First, inspired by the artificial bee colony algorithm in evolutionary computation, we propose an architecture evolution method, namely ABC-Arc. This method optimizes network topology by systematically removing redundant neural pathways. Second, a dynamic post-training quantization (DynPTQ) strategy is developed for the evolved SpikeLM, facilitating the conversion of floating-point parameters to lower-bit precision without requiring model retraining. By combining these two methods, EQ-SpikeLM significantly reduces storage and computational demands while preserving model performance. Experimental evaluation on the GLUE benchmark demonstrates EQ-SpikeLM’s ability to maintain performance equivalent to its uncompressed counterpart, with a substantial reduction in both model size and power consumption. These results position EQ-SpikeLM as a viable approach for deploying large language models in resource-constrained edge computing scenarios.

The Fourier transform (FT) stands as a fundamental tool in modern signal processing with widespread applications across various scientific and engineering fields. Therefore, there remains a need for continued research efforts to devise energy-efficient implementations of the FT. Due to their inherent energy efficiency, biologically plausible spiking neural networks (SNNs) emerge as a promising alternative solution. However, current SNN implementations of the FT suffer from two key shortcomings, namely, high latency and reduced accuracy. In this article, we analyze the underlying causes of these limitations and highlight deficiencies in the existing spike-based encoding mechanisms and spiking neuron models. We then propose a new SNN-based FT (SNN-FT) based on a logarithmically polarized time-to-first-spike (TTFS) encoding method (called LP-TTFS) along with a novel piecewise spiking neuron (PTSN) model based on ternary spikes (referred to as PTSN). The resulting SNN-FT is mathematically equivalent to the conventional FT and demonstrates superior performance in accuracy as well as reduced latency. We assess the performance of the proposed SNN-FT alternative through extensive experiments on FT-based applications, such as radar and audio signal processing, and the obtained results demonstrate the efficacy of SNN-FT and its superiority over the existing approaches. This study unveils a novel energy-efficient neuromorphic computing technique with great potential for FT applications across diverse scientific and engineering domains.

Binary Spiking Neural Networks (BSNNs) inherit the event-driven paradigm of SNNs, while also adopting the reduced storage burden of binarization techniques. These distinct advantages grant BSNNs lightweight and energy-efficient characteristics, rendering them ideal for deployment on resource-constrained edge devices. However, due to the binary synaptic weights and non-differentiable spike function, effectively training BSNNs remains an open question. In this paper, we conduct an in-depth analysis of the challenge for BSNN learning, namely the frequent weight sign flipping problem. To mitigate this issue, we propose an Adaptive Gradient Modulation Mechanism (AGMM), which is designed to reduce the frequency of weight sign flipping by adaptively adjusting the gradients during the learning process. The proposed AGMM can enable BSNNs to achieve faster convergence speed and higher accuracy, effectively narrowing the gap between BSNNs and their full-precision equivalents. We validate AGMM on both static and neuromorphic datasets, and results indicate that it achieves state-of-the-art results among BSNNs. This work substantially reduces storage demands and enhances SNNs' inherent energy efficiency, making them highly feasible for resource-constrained environments.

The combination of Spiking Neural Networks (SNNs) and Vision Transformers (ViTs) holds potential for achieving both energy efficiency and high performance, particularly suitable for edge vision applications. However, a significant performance gap still exists between SNN-based ViTs and their ANN counterparts. Here, we first analyze why SNN-based ViTs suffer from limited performance and identify a mismatch between the vanilla self-attention mechanism and spatio-temporal spike trains. This mismatch results in degraded spatial relevance and limited temporal interactions. To address these issues, we draw inspiration from biological saccadic attention mechanisms and introduce an innovative Saccadic Spike Self-Attention (SSSA) method. Specifically, in the spatial domain, SSSA employs a novel spike distribution-based method to effectively assess the relevance between Query and Key pairs in SNN-based ViTs. Temporally, SSSA employs a saccadic interaction module that dynamically focuses on selected visual areas at each timestep and significantly enhances whole scene understanding through temporal interactions. Building on the SSSA mechanism, we develop a SNN-based Vision Transformer (SNN-ViT). Extensive experiments across various visual tasks demonstrate that SNN-ViT achieves state-of-the-art performance with linear computational complexity. The effectiveness and efficiency of the SNN-ViT highlight its potential for power-critical edge vision applications.

Brain-inspired Spiking Neural Networks (SNNs) leverage sparse spikes to encode information and operate in an asynchronous event-driven manner, offering a highly energy-efficient paradigm for machine intelligence. However, the current SNN community focuses primarily on performance improvement by developing large-scale models, which limits the applicability of SNNs in resource-limited edge devices. In this paper, we propose a hardware-friendly and lightweight SNN, aimed at effectively deploying high-performance SNN in resource-limited scenarios. Specifically, we first develop a baseline model that integrates uniform quantization and structured pruning, called QP-SNN baseline. While this baseline significantly reduces storage demands and computational costs, it suffers from performance decline. To address this, we conduct an in-depth analysis of the challenges in quantization and pruning that lead to performance degradation and propose solutions to enhance the baseline's performance. For weight quantization, we propose a weight rescaling strategy that utilizes bit width more effectively to enhance the model's representation capability. For structured pruning, we propose a novel pruning criterion using the singular value of spatiotemporal spike activities to enable more accurate removal of redundant kernels. Extensive experiments demonstrate that integrating two proposed methods into the baseline allows QP-SNN to achieve state-of-the-art performance and efficiency, underscoring its potential for enhancing SNN deployment in edge intelligence computing.

Binary Spiking Neural Networks (BSNNs) offer promising efficiency advantages for resource-constrained computing. However, their training algorithms often require substantial memory overhead due to latent weights storage and temporal processing requirements. To address this issue, we propose Binary Spiking Online (BSO) optimization algorithm, a novel online training algorithm that significantly reduces training memory. BSO directly updates weights through flip signals under the online training framework. These signals are triggered when the product of gradient momentum and weights exceeds a threshold, eliminating the need for latent weights during training. To enhance performance, we propose T-BSO, a temporal-aware variant that leverages the inherent temporal dynamics of BSNNs by capturing gradient information across time steps for adaptive threshold adjustment. Theoretical analysis establishes convergence guarantees for both BSO and T-BSO, with formal regret bounds characterizing their convergence rates. Extensive experiments demonstrate that both BSO and T-BSO achieve superior optimization performance compared to existing training methods for BSNNs. The codes are available at https://github.com/hamingsi/BSO.

Transformer-based Spiking Neural Networks (SNNs) introduce a novel event-driven self-attention paradigm that combines the high performance of Transformers with the energy efficiency of SNNs. However, the larger model size and increased computational demands of the Transformer structure limit their practicality in resource-constrained scenarios. In this paper, we integrate binarization techniques into Transformer-based SNNs and propose the Binary Event-Driven Spiking Transformer, i.e. BESTformer. The proposed BESTformer can significantly reduce storage and computational demands by representing weights and attention maps with a mere 1-bit. However, BESTformer suffers from a severe performance drop from its full-precision counterpart due to the limited representation capability of binarization. To address this issue, we propose a Coupled Information Enhancement (CIE) method, which consists of a reversible framework and information enhancement distillation. By maximizing the mutual information between the binary model and its full-precision counterpart, the CIE method effectively mitigates the performance degradation of the BESTformer. Extensive experiments on static and neuromorphic datasets demonstrate that our method achieves superior performance to other binary SNNs, showcasing its potential as a compact yet high-performance model for resource-limited edge devices. The repository of this paper is available at https://github.com/CaoHLin/BESTFormer.

Spiking Neural Networks (SNNs) offer an energy-efficient paradigm for machine intelligence, but their continued scaling poses challenges for resource-limited deployment. Despite recent advances in binary SNNs, the storage and computational demands remain substantial for large-scale networks. To further explore the compression and acceleration potential of SNNs, we propose Sub-bit Spiking Neural Networks (S$^2$NNs) that represent weights with less than one bit. Specifically, we first establish an S$^2$NN baseline by leveraging the clustering patterns of kernels in well-trained binary SNNs. This baseline is highly efficient but suffers from \textit{outlier-induced codeword selection bias} during training. To mitigate this issue, we propose an \textit{outlier-aware sub-bit weight quantization} (OS-Quant) method, which optimizes codeword selection by identifying and adaptively scaling outliers. Furthermore, we propose a \textit{membrane potential-based feature distillation} (MPFD) method, improving the performance of highly compressed S$^2$NN via more precise guidance from a teacher model. Extensive results on vision reveal that S$^2$NN outperforms existing quantized SNNs in both performance and efficiency, making it promising for edge computing applications.

The brain is able to acquire and store memories of everyday experiences in real-time. It can also selectively forget information to facilitate memory updating. However, our understanding of the underlying mechanisms and coordination of these processes within the brain remains limited. However, no existing artificial intelligence models have yet matched human-level capabilities in terms of memory storage and retrieval. This study introduces a brain-inspired spiking neural model that integrates the learning and forgetting processes of sequential memory. The proposed model closely mimics the distributed and sparse temporal coding observed in the biological neural system. It employs one-shot online learning for memory formation and uses biologically plausible mechanisms of neural oscillation and phase precession to retrieve memorized sequences reliably. In addition, an active forgetting mechanism is integrated into the spiking neural model, enabling memory removal, flexibility, and updating. The proposed memory model not only enhances our understanding of human memory processes but also provides a robust framework for addressing temporal modeling tasks.

Stock market is an open marketplace for the creation, acquisition, and exchange of stocks that trade over the counter or on a stock exchange, this market accommodates billions of transactions. Stock market manipulation occurs when dealers attempt to fraudulently raise or lower a stock price to personal benefit. Pump and dump is a common manipulation technique that is frequently used to artificially boost and collapse stock prices and manually determining such activity has proven to be cumbersome. Machine learning and deep learning models have been utilized to recognize a range of stock manipulation scenarios; however, they lack the ability to model dynamically in continuous real time and trend identification gets hindered due to significant noise-to-signal ratio. Hence, this work focused on the viability of Spiking Neural Networks (SNNs) to naturally adapt and manage non-linear and temporal based input data that classical neural networks struggle with. A feed-forward network of Leaky-Integrate and Fire (LIF) neurons is proposed for stock market manipulation detection. The data employed in this research was obtained from the LOBSTER project and the Bloomberg Newcastle Business School trading room. To find the ideal network design and encoding strategy for this task, extensive experiments are conducted and experimental results and their comparison against existing models revealed that the proposed method outperforms the chosen benchmark models and is more successful at identifying patterns of stock price manipulation.

Stock price manipulation refers to deceptive traders' practices which aim to influence the normal market behaviour in order to make illicit profit at the expense of other genuine market participants. Such manipulation undermines investors’ confidence in the financial market and damages its efficiency and integrity. There remains a growing need for developing robust anomaly detection methods capable of reliably identifying increasingly sophisticated manipulation attempts. Inspired by the recent success of deep learning in computer vision, we propose a novel approach for stock market manipulation detection that leverages the combined power of Recurrence Plots (RP) and beta Variational Autoencoders (beta-VAEs). The proposed approach first splits stock price time series data into overlapping temporal windows, each of which is transformed into a colour image using Recurrence Plots (RP). Then, a beta-VAE network composed of two Convolutional Neural Networks (CNNs) is trained on these images derived from normal market activity. The resulting model effectively learns the inherent characteristics of normal (i.e legitimate) trading behaviour. Finally, the mean square error between the original images and the reconstructed ones is used as the manipulation (i.e anomaly) detection score measure for flagging significant deviations indicative of potential manipulation. The efficacy of the proposed approach is rigorously validated on 1-level tick data obtained from the LOBSTER project. Evaluation results demonstrate superior manipulation detection performance which is evidenced by a highly promising area under the ROC curve (AUC). The robustness of this performance can be attributed to the beta-VAE's ability to extract pertinent features of normal market behavior from the proposed RP-generated 2D representations of stock price data. This novel framework paves the way for the development of more effective market abuse detection systems which contribute to a safer, fairer and more transparent financial ecosystem for both investors and regulators alike.

Brain-inspired Spiking Neural Networks (SNNs) leverage sparse spikes to represent information and process them in an asynchronous event-driven manner, offering an energy-efficient paradigm for the next generation of machine intelligence. However, the current focus within the SNN community prioritizes accuracy optimization through the development of large-scale models, limiting their viability in resource-constrained and low-power edge devices. To address this challenge, we introduce a lightweight and hardware-friendly Quantized SNN (Q-SNN) that applies quantization to both synaptic weights and membrane potentials. By significantly compressing these two key elements, the proposed Q-SNNs substantially reduce both memory usage and computational complexity. Moreover, to prevent the performance degradation caused by this compression, we present a new Weight-Spike Dual Regulation (WS-DR) method inspired by information entropy theory. Experimental evaluations on various datasets, including static and neuromorphic, demonstrate that our Q-SNNs outperform existing methods in terms of both model size and accuracy. These state-of-the-art results in efficiency and efficacy suggest that the proposed method can significantly improve edge intelligent computing.

Biological systems possess remarkable sound source localization (SSL) capabilities that are critical for survival in complex environments. This ability arises from the collaboration between the auditory periphery, which encodes sound as precisely timed spikes, and the auditory cortex, which performs spike-based computations. Inspired by these biological mechanisms, we propose a novel neuromorphic SSL framework that integrates spike-based neural encoding and computation. The framework employs Resonate-and-Fire (RF) neurons with a phase-locking coding (RF-PLC) method to achieve energy-efficient audio processing. The RF-PLC method leverages the resonance properties of RF neurons to efficiently convert audio signals to time-frequency representation and encode interaural time difference (ITD) cues into discriminative spike patterns. In addition, biological adaptations like frequency band selectivity and short-term memory effectively filter out many environmental noises, enhancing SSL capabilities in real-world settings. Inspired by these adaptations, we propose a spike-driven multi-auditory attention (MAA) module that significantly improves both the accuracy and robustness of the proposed SSL framework. Extensive experimentation demonstrates that our SSL framework achieves state-of-the-art accuracy in SSL tasks. Furthermore, it shows exceptional noise robustness and maintains high accuracy even at very low signal-to-noise ratios. By mimicking biological hearing, this neuromorphic approach contributes to the development of high-performance and explainable artificial intelligence systems capable of superior performance in real-world environments.

The domain of image classification has been seen to be dominated by high-performing deep-learning (DL) architectures. However, the success of this field, as seen over the past decade, has resulted in the complexity of modern methodologies scaling exponentially, commonly requiring millions of parameters. Quantum computing (QC) is an active area of research aimed toward greatly reducing problems of complexity faced in classical computing. With growing interest toward quantum machine learning (QML) for applications of image classification, many proposed algorithms require usage of numerous qubits. In the noisy intermediate-scale quantum (NISQ) era, these circuits may not always be feasible to execute effectively; therefore, we should aim to use each qubit as effectively and efficiently as possible, before adding additional qubits. This article proposes a new single-qubit-based deep quantum neural network for image classification that mimics traditional convolutional neural network (CNN) techniques, resulting in a reduced number of parameters compared with previous works. Our aim is to prove the concept of the initial proposal by demonstrating classification performance of the single-qubit-based architecture, as well as to provide a tested foundation for further development. To demonstrate this, our experiments were conducted using various datasets including MNIST, Fashion-MNIST, and ORL face datasets. To further our proposal in the context of the NISQ era, our experiments were intentionally conducted in noisy simulation environments. Initial test results appear promising, with classification accuracies of 94.6%, 89.5%, and 82.5% achieved on the subsets of MNIST, FMNIST, and ORL face datasets, respectively. In addition, proposals for further investigation and development were considered, where it is hoped that these initial results can be improved.

Spiking Neural Networks (SNNs) offer a highly promising computing paradigm due to their biological plausibility, exceptional spatiotemporal information processing capability and low power consumption. As a temporal encoding scheme for SNNs, Time-To-First-Spike (TTFS) encodes information using the timing of a single spike, which allows spiking neurons to transmit information through sparse spike trains and results in lower power consumption and higher computational efficiency compared to traditional rate-based encoding counterparts. However, despite the advantages of the TTFS encoding scheme, the effective and efficient training of TTFS-based deep SNNs remains a significant and open research problem. In this work, we first examine the factors underlying the limitations of applying existing TTFS-based learning algorithms to deep SNNs. Specifically, we investigate issues related to over-sparsity of spikes and the complexity of finding the ‘causal set'. We then propose a simple yet efficient dynamic firing threshold (DFT) mechanism for spiking neurons to address these issues. Building upon the proposed DFT mechanism, we further introduce a novel direct training algorithm for TTFS-based deep SNNs, called DTA-TTFS. This method utilizes event-driven processing and spike timing to enable efficient learning of deep SNNs. The proposed training method was validated on the image classification task and experimental results clearly demonstrate that our proposed method achieves state-of-the-art accuracy in comparison to existing TTFS-based learning algorithms, while maintaining high levels of sparsity and energy efficiency on neuromorphic inference accelerator.

Palm vein images usually suffer from low contrast due to skin surface scattering the radiance of NIR light and image sensor limitations, hence require employing various techniques to enhance the contrast of the image prior to feature extraction. This paper presents a novel image enhancement method referred to as Multiple Overlapping Tiles (MOT) which adaptively stretches the local contrast of palm vein images using multiple layers of overlapping image tiles. The experiments conducted on the CASIA palm vein image dataset demonstrate that the MOT method retains the finer subspace details of vein images which allows excellent feature detection and matching with SIFT and RootSIFT features. Results on existing palm vein recognition systems demonstrate that the proposed MOT method delivers lower EER values outperforming other existing palm vein image enhancement methods.

Abstract Steganography is the art of secret communication and steganalysis is the art of detecting the hidden messages embedded in digital media covers. One of the covers that is gaining interest in the field is video. Presently, the global IP video traffic forms the major part of all consumer Internet traffic. It is also gaining attention in the field of digital forensics and homeland security in which threats of covert communications hold serious consequences. Thus, steganography technicians will prefer video to other types of covers like audio files, still images, or texts. Moreover, video steganography will be of more interest because it provides more concealing capacity. Contrariwise, investigation in video steganalysis methods does not seem to follow the momentum even if law enforcement agencies and governments around the world support and encourage investigation in this field. In this paper, the authors review the most important methods used so far in video steganalysis and sketch the future trends. To the best of the authors’ knowledge this is the most comprehensive review of video steganalysis produced so far.

Spiking neural networks (SNNs) use spatiotemporal spike patterns to represent and transmit information, which are not only biologically realistic but also suitable for ultralow-power event-driven neuromorphic implementation. Just like other deep learning techniques, deep SNNs (DeepSNNs) benefit from the deep architecture. However, the training of DeepSNNs is not straightforward because the well-studied error backpropagation (BP) algorithm is not directly applicable. In this article, we first establish an understanding as to why error BP does not work well in DeepSNNs. We then propose a simple yet efficient rectified linear postsynaptic potential function (ReL-PSP) for spiking neurons and a spike-timing-dependent BP (STDBP) learning algorithm for DeepSNNs where the timing of individual spikes is used to convey information (temporal coding), and learning (BP) is performed based on spike timing in an event-driven manner. We show that DeepSNNs trained with the proposed single spike time-based learning algorithm can achieve the state-of-the-art classification accuracy. Furthermore, by utilizing the trained model parameters obtained from the proposed STDBP learning algorithm, we demonstrate ultralow-power inference operations on a recently proposed neuromorphic inference accelerator. The experimental results also show that the neuromorphic hardware consumes 0.751 mW of the total power consumption and achieves a low latency of 47.71 ms to classify an image from the Modified National Institute of Standards and Technology (MNIST) dataset. Overall, this work investigates the contribution of spike timing dynamics for information encoding, synaptic plasticity, and decision-making, providing a new perspective to the design of future DeepSNNs and neuromorphic hardware.

Stock price manipulation, a major problem in capital markets surveillance, uses illegitimate means to influence the price of traded stocks in order to reap illicit profit. Most of the existing attempts to detect such manipulations have either relied upon annotated trading data, using supervised methods, or have been restricted to detecting a specific manipulation scheme. There have been a few unsupervised algorithms focusing on general detection yet none of them explored the innate affinity among the stock trades, be it normal or manipulative. This paper proposes a fully unsupervised model based on the idea of learning the relationship among stock prices in the form of an affinity matrix. The proposed affinity matrix based features are used to train an under-fitting autoencoder in order to learn an efficient representation of the normal stock prices. A kernel density estimate of the normal trading data is used as the reconstruction error of the autoencoder. During the detection phase, the normal dataset has been injected with synthetic manipulative trades. A kernel density estimation based clustering technique is then used to detect manipulative trades based on their autoencoder representation. The proposed approach is validated on benchmark stock price data from the LOBSTER project and the obtained results show dramatic improvements in the detection performance over existing price manipulation detection techniques.

Summary Image segmentation plays a crucial role in many image processing and understanding applications. Despite the huge number of proposed image segmentation techniques, accurate segmentation remains a significant challenge in image analysis. This article investigates the viability of using echo state network (ESN), a biologically inspired recurrent neural network, as features extractor for efficient color image segmentation. First, an ensemble of initial pixel features is extracted from the original images and injected into the ESN reservoir. Second, the internal activations of the reservoir neurons are used as new pixel features. Third, the new features are classified using a feed forward neural network as a readout layer for the ESN. The quality of the pixel features produced by the ESN is evaluated through extensive series of experiments conducted on real world image datasets. The optimal operating range of different ESN setup parameters for producing competitive quality features is identified. The performance of the proposed ESN‐based framework is also evaluated on a domain‐specific application, namely, blood vessel segmentation in retinal images where experiments are conducted on the widely used digital retinal images for vessel extraction (DRIVE) dataset. The obtained results demonstrate that the proposed method outperforms state‐of‐the‐art general segmentation techniques in terms of performance with an F‐score of 0.92 ± 0.003 on the segmentation evaluation dataset. In addition, the proposed method achieves a comparable segmentation accuracy (0.9470) comparing with reported techniques of segmentation of blood vessels in images of retina and outperform them in terms of processing time. The average time required by our technique to segment one retinal image from DRIVE dataset is 8 seconds. Furthermore, empirically derived guidelines are proposed for adequately setting the ESN parameters for effective color image segmentation.

Market manipulation is the act of artificially influencing the price of a security to make profit through illegitimate schemes. It is evident from the literature that only a handful of methods had been proposed for stock market manipulation detection. Most of those methods either used supervised training or focused only on specific manipulation schemes. This paper introduces a semi-supervised learning method based on a hybridization of an altered dendritic cell immune system inspired approach and Kernel Density Estimation based clustering technique. Dendritic Cell Algorithm (DCA) mimics the human immune system in data processing using the danger theory model for anomaly detection. An important advantage of the proposed approach is that the DCA is adapted for scaling down the dimension of the input data set to a set of only three outputs that are then clustered using KDE clustering. This avoids the need for assigning different threshold parameters as in a conventional DCA, hence automating the detection process. Another important advantage is that supervised training is not required for signal categorization during the preprocessing phase of DCA. The proposed approach is validated on Level 1 stock price tick data obtained from the LOBSTER project which contains highly volatile and high frequency trading (HFT) time series. The considered manipulation schemes are Pump and Dump and Gouging or Spoof trading. The proposed approach is benchmarked against existing stock market manipulation detection approaches as well as existing anomaly detection techniques based on KNN, OCSVM, PCA and k-means. The obtained results show substantial improvements in terms of the area under the ROC curve (AUC) and the false alarm rate.

Spiking neurons are becoming increasingly popular owing to their biological plausibility and promising computational properties. Unlike traditional rate-based neural models, spiking neurons encode information in the temporal patterns of the transmitted spike trains, which makes them more suitable for processing spatiotemporal information. One of the fundamental computations of spiking neurons is to transform streams of input spike trains into precisely timed firing activity. However, the existing learning methods, used to realize such computation, often result in relatively low accuracy performance and poor robustness to noise. In order to address these limitations, we propose a novel highly effective and robust membrane potential-driven supervised learning (MemPo-Learn) method, which enables the trained neurons to generate desired spike trains with higher precision, higher efficiency, and better noise robustness than the current state-of-the-art spiking neuron learning methods. While the traditional spike-driven learning methods use an error function based on the difference between the actual and desired output spike trains, the proposed MemPo-Learn method employs an error function based on the difference between the output neuron membrane potential and its firing threshold. The efficiency of the proposed learning method is further improved through the introduction of an adaptive strategy, called skip scan training strategy, that selectively identifies the time steps when to apply weight adjustment. The proposed strategy enables the MemPo-Learn method to effectively and efficiently learn the desired output spike train even when much smaller time steps are used. In addition, the learning rule of MemPo-Learn is improved further to help mitigate the impact of the input noise on the timing accuracy and reliability of the neuron firing dynamics. The proposed learning method is thoroughly evaluated on synthetic data and is further demonstrated on real-world classification tasks. Experimental results show that the proposed method can achieve high learning accuracy with a significant improvement in learning time and better robustness to different types of noise.

There is a biological evidence to prove information is coded through precise timing of spikes in the brain. However, training a population of spiking neurons in a multilayer network to fire at multiple precise times remains a challenging task. Delay learning and the effect of a delay on weight learning in a spiking neural network (SNN) have not been investigated thoroughly. This paper proposes a novel biologically plausible supervised learning algorithm for learning precisely timed multiple spikes in a multilayer SNNs. Based on the spike-timing-dependent plasticity learning rule, the proposed learning method trains an SNN through the synergy between weight and delay learning. The weights of the hidden and output neurons are adjusted in parallel. The proposed learning method captures the contribution of synaptic delays to the learning of synaptic weights. Interaction between different layers of the network is realized through biofeedback signals sent by the output neurons. The trained SNN is used for the classification of spatiotemporal input patterns. The proposed learning method also trains the spiking network not to fire spikes at undesired times which contribute to misclassification. Experimental evaluation on benchmark data sets from the UCI machine learning repository shows that the proposed method has comparable results with classical rate-based methods such as deep belief network and the autoencoder models. Moreover, the proposed method can achieve higher classification accuracies than single layer and a similar multilayer SNN.

The creation of a predictive system that correctly forecasts future changes of a stock price is crucial for investment management and algorithmic trading. The use of technical analysis for financial forecasting has been successfully employed by many researchers. Input window length is a time frame parameter required to be set when calculating many technical indicators. This study explores how the performance of the predictive system depends on a combination of a forecast horizon and an input window length for forecasting variable horizons. Technical indicators are used as input features for machine learning algorithms to forecast future directions of stock price movements. The dataset consists of ten years daily price time series for fifty stocks. The highest prediction performance is observed when the input window length is approximately equal to the forecast horizon. This novel pattern is studied using multiple performance metrics: prediction accuracy, winning rate, return per trade and Sharpe ratio.

This paper presents an enhanced rank-order-based learning algorithm, called SpikeTemp, for spiking neural networks (SNNs) with a dynamically adaptive structure. The trained feed-forward SNN consists of two layers of spiking neurons: 1) an encoding layer which temporally encodes real-valued features into spatio-temporal spike patterns and 2) an output layer of dynamically grown neurons which perform spatio-temporal classification. Both Gaussian receptive fields and square cosine population encoding schemes are employed to encode real-valued features into spatio-temporal spike patterns. Unlike the rank-order-based learning approach, SpikeTemp uses the precise times of the incoming spikes for adjusting the synaptic weights such that early spikes result in a large weight change and late spikes lead to a smaller weight change. This removes the need to rank all the incoming spikes and, thus, reduces the computational cost of SpikeTemp. The proposed SpikeTemp algorithm is demonstrated on several benchmark data sets and on an image recognition task. The results show that SpikeTemp can achieve better classification performance and is much faster than the existing rank-order-based learning approach. In addition, the number of output neurons is much smaller when the square cosine encoding scheme is employed. Furthermore, SpikeTemp is benchmarked against a selection of existing machine learning algorithms, and the results demonstrate the ability of SpikeTemp to classify different data sets after just one presentation of the training samples with comparable classification performance.

A system of interacting agents is, by definition, very demanding in terms of computational resources. Although multi-agent systems have been used to solve complex problems in many areas, it is usually very difficult to perform large-scale simulations in their targeted serial computing platforms. Reconfigurable hardware, in particular Field Programmable Gate Arrays devices, have been successfully used in High Performance Computing applications due to their inherent flexibility, data parallelism, and algorithm acceleration capabilities. Indeed, reconfigurable hardware seems to be the next logical step in the agency paradigm, but only a few attempts have been successful in implementing multi-agent systems in these platforms. This article discusses the problem of inter-agent communications in Field Programmable Gate Arrays. It proposes a Network-on-Chip in a hierarchical star topology to enable agents’ transactions through message broadcasting using the Open Core Protocol as an interface between hardware modules. A customizable router microarchitecture is described and a multi-agent system is created to simulate and analyse message exchanges in a generic heavy traffic load agent-based application. Experiments have shown a throughput of 1.6Gbps per port at 100MHz without packet loss and seamless scalability characteristics.

Image segmentation refers to the process of dividing an image into multiple regions which represent meaningful areas. Image segmentation is an essential step for most image analysis tasks such as object recognition and tracking, pattern recognition, content-based image retrieval, etc. In recent years, a large number of image segmentation algorithms have been developed, but achieving accurate segmentation still remains a challenging task. Recently, reservoir computing (RC) has drawn much attention in machine learning as a new model of recurrent neural networks (RNN). Echo State Network (ESN) represents one efficient realization of RC, which is initially designed to facilitate learning in Recurrent Neural Networks. In this paper we investigate the viability of ESN as feature extractor for pixel classification based colour image segmentation. Extensive experiments are conducted on real world colour image datasets and the global ESN reservoir parameters are varied to identify their operating ranges that allow the use of the reservoir nodes internal activations as new pixel features for the colour image segmentation task. A simple feed forward neural network is used to realize the ESN readout function and classify these new features. The experimental results show that the proposed method achieves high performance image segmentation comparing with state-of-the-art techniques. In addition, a set of empirically derived guidelines for setting the reservoir global parameters are proposed.

A wash trade refers to the illegal activities of traders who utilize carefully designed limit orders to manually increase the trading volumes for creating a false impression of an active market. As one of the primary formats of market abuse, a wash trade can be extremely damaging to the proper functioning and integrity of capital markets. The existing work focuses on collusive clique detections based on certain assumptions of trading behaviors. Effective approaches for analyzing and detecting wash trade in a real-life market have yet to be developed. This paper analyzes and conceptualizes the basic structures of the trading collusion in a wash trade by using a directed graph of traders. A novel method is then proposed to detect the potential wash trade activities involved in a financial instrument by first recognizing the suspiciously matched orders and then further identifying the collusions among the traders who submit such orders. Both steps are formulated as a simplified form of the knapsack problem, which can be solved by dynamic programming approaches. The proposed approach is evaluated on seven stock data sets from the NASDAQ and the London Stock Exchange. The experimental results show that the proposed approach can effectively detect all primary wash trade scenarios across the selected data sets.

Face detection is one of the most active research areas in computer vision. Despite the well documented success of classical machine learning techniques in controlled situations, face detection in completely uncontrolled settings remains a difficult task. Recent progress with bio-inspired approaches have addressed challenging areas of invariance including scale, occlusion and illumination issues, but there remains a lack of concentrated effort into truly multi-view detection of faces in different poses and orientations. This paper introduces a novel strategy to address this through the enhanced implementation of a hierarchical bio-inspired HMAX framework using spiking neurons that implements feature extraction with unsupervised STDP. A multiple trial training scheme is introduced to train separate pools of neurons on different face poses. The trained neurons are then processed by an additional STDP mechanism to generate a streamlined repository of broadly tuned multi-view neurons. Experimental results demonstrate that the new system achieves robust invariant detection of in-plane and out-of-plane rotated faces with single face per image datasets. In addition, extending the multi-view system by introducing lateral inhibition between merged pools of multi-view face detecting neurons, results in a single model that is able to achieve simultaneous face detection and accurate face pose estimation.

Accurate forecasting of upcoming trends in the capital markets is extremely important for algorithmic trading and investment management. Before making a trading decision, investors estimate the probability that a certain news item will influence the market based on the available information. Speculation among traders is often caused by the release of a breaking news article and results in price movements. Publications of news articles influence the market state that makes them a powerful source of data in financial forecasting. Recently, researchers have developed trend and price prediction models based on information extracted from news articles. However, to date no previous research that investigates the advantages of using news articles with different levels of relevance to the target stock has been conducted. This research study uses the multiple kernel learning technique to effectively combine information extracted from stock-specific and sub-industry-specific news articles for prediction of an upcoming price movement. News articles are divided into these two categories based on their relevance to a targeted stock and analyzed by separate kernels. The experimental results show that utilizing two categories of news improves the prediction accuracy in comparison with methods based on a single news category.

Spikes are an important part of information transmission between neurons in the biological brain. Biological evidence shows that information is carried in the timing of individual action potentials, rather than only the firing rate. Spiking neural networks are devised to capture more biological characteristics of the brain to construct more powerful intelligent systems. In this paper, we extend our newly proposed supervised learning algorithm called DL-ReSuMe (Delay Learning Remote Supervised Method) to train multiple neurons to classify spatiotemporal spiking patterns. In this method, a number of neurons instead of a single neuron is trained to perform the classification task. The simulation results show that a population of neurons has significantly higher processing ability compared to a single neuron. It is also shown that the performance of Multi-DL-ReSuMe (Multiple DL-ReSuMe) is increased when the number of desired spikes is increased in the desired spike trains to an appropriate number.

This paper presents a novel RBF-like fast dynamically Evolving Spiking Neural classifier (ESNC). The trained feed-forward SNN consists of three layers of spiking neurons: an encoding layer which temporally encodes real valued features into spatio-temporal spike patterns, a hidden layer of dynamically grown and pruned neurons which perform spatiotemporal clustering, and an evolving output layer for classification. Both the structure and weights of the SNN are learned dynamically through a combination of unsupervised and supervised learning paradigms. An unsupervised clustering method is implemented by the hidden layer for adjusting the synaptic weights of the hidden neurons afferent connections. The centre of each hidden RBF neuron is represented by a vector of temporal distances between the first spike of the hidden neuron and the presynaptic spikes. In addition, strategies are proposed to adjust the structure of the hidden and output layers as inputs are presented to the SNN, and classification at the output layer is achieved through supervised learning where a learning window is used to adjust the weights of the output neurons afferent connections. The proposed learning algorithm is demonstrated on several benchmark datasets from the UCL machine learning repository. The results show comparable performance with existing machine learning algorithms and demonstrate the ability of the proposed algorithm to learn incoming data samples in a hybrid way and in one epoch only.

Price manipulation refers to the activities of those traders who use carefully designed trading behaviors to manually push up or down the underlying equity prices for making profits. With increasing volumes and frequency of trading, price manipulation can be extremely damaging to the proper functioning and integrity of capital markets. The existing literature focuses on either empirical studies of market abuse cases or analysis of particular manipulation types based on certain assumptions. Effective approaches for analyzing and detecting price manipulation in real time are yet to be developed. This paper proposes a novel approach, called adaptive hidden Markov model with anomaly states (AHMMAS) for modeling and detecting price manipulation activities. Together with wavelet transformations and gradients as the feature extraction methods, the AHMMAS model caters to price manipulation detection and basic manipulation type recognition. The evaluation experiments conducted on seven stock tick data from NASDAQ and the London Stock Exchange and 10 simulated stock prices by stochastic differential equation show that the proposed AHMMAS model can effectively detect price manipulation patterns and outperforms the selected benchmark models.

This paper presents a level set boundary description (LSBD) approach for novelty detection that treats the nonlinear boundary directly in the input space. The proposed approach consists of level set function (LSF) construction, boundary evolution, and termination of the training process. It employs kernel density estimation to construct the LSF of the initial boundary for the training data set. Then, a sign of the LSF-based algorithm is proposed to evolve the boundary and make it fit more tightly in the data distribution. The training process terminates when an expected fraction of rejected normal data is reached. The evolution process utilizes the signs of the LSF values at all training data points to decide whether to expand or shrink the boundary. Extensive experiments are conducted on benchmark data sets to evaluate the proposed LSBD method and compare it against four representative novelty detection methods. The experimental results demonstrate that the novelty detector modeled with the proposed LSBD can effectively detect anomalies.

Recent research has shown the potential capability of spiking neural networks (SNNs) to model complex information processing in the brain. There is biological evidence to prove the use of the precise timing of spikes for information coding. However, the exact learning mechanism in which the neuron is trained to fire at precise times remains an open problem. The majority of the existing learning methods for SNNs are based on weight adjustment. However, there is also biological evidence that the synaptic delay is not constant. In this paper, a learning method for spiking neurons, called delay learning remote supervised method (DL-ReSuMe), is proposed to merge the delay shift approach and ReSuMe-based weight adjustment to enhance the learning performance. DL-ReSuMe uses more biologically plausible properties, such as delay learning, and needs less weight adjustment than ReSuMe. Simulation results have shown that the proposed DL-ReSuMe approach achieves learning accuracy and learning speed improvements compared with ReSuMe.

Market abuse has attracted much attention from financial regulators around the world but it is difficult to fully prevent. One of the reasons is the lack of thoroughly studies of the market abuse strategies and the corresponding effective market abuse approaches. In this paper, the strategies of reported price manipulation cases are analysed as well as the related empirical studies. A transformation is then defined to convert the time-varying financial trading data into pseudo-stationary time series, where machine learning algorithms can be easily applied to the detection of the price manipulation. The evaluation experiments conducted on four stocks from NASDAQ show a promising improved performance for effectively detecting such manipulation cases.

Wash trade refers to the activities of traders who utilise deliberately designed collusive transactions to increase the trading volumes for creating active market impression. Wash trade can be damaging to the proper functioning and integrity of capital markets. Existing work focuses on collusive clique detections based on certain assumptions of trading behaviours. Effective approaches for analysing and detecting wash trade in a real-life market have yet to be developed. This paper proposes a new analysis approach for abstracting the basic structures of wash trade based on the network topology theory and a novel approach for detecting wash trade activities. The evaluation experiments conducted on four NASDAQ stocks suggest that wash trade actions can be effectively identified based on the proposed algorithm.

Electronic trading in global markets and exchanges requires sophisticated communication and data management systems. Novel computational infrastructures and trading strategies are required to support the massive amount of incoming streaming data, where the main problem is in latency management. Multi-agent Systems have been recognized as a promising solution to address complex problems in many areas such as biology, social sciences and financial markets and may provide powerful and flexible solutions for implementing trading engines. In addition, reconfigurable hardware based on Field Programmable Gate Arrays (FPGAs) offers many important performance benefits over software implementations, such as reducing decision making latency and high-throughput data processing. Robust and scalable trading engines can be developed by leveraging the benefits of reconfigurable FPGA platforms. This paper presents a multi-agent architecture in reconfigurable hardware for financial applications and the implementation of a trading engine for pre-trade analysis as a validation scenario. Performance results show that calculation of technical indicators and trading strategy evaluation to generate trading signals with a latency of 550 ns is achievable.

In recent years, machine learning algorithms have become increasingly popular in financial forecasting. Their flexible, data-driven nature makes them ideal candidates for dealing with complex financial data. This paper investigates the effectiveness of a number of machine learning algorithms, and combinations of these algorithms, at generating one-step ahead forecasts of a number of financial time series. We find that hybrid models consisting of a linear statistical model and a nonlinear machine learning algorithm are effective at forecasting future values of the series, particularly in terms of the future direction of the series.

Accurate forecasting of directional changes in stock prices is important for algorithmic trading and investment management. Technical analysis has been successfully used in financial forecasting and recently researchers have explored the optimization of parameters for technical indicators. This study investigates the relationship between the window size used for calculating technical indicators and the accuracy of one-step-ahead (variable steps) forecasting. The directions of the future price movements are predicted using technical analysis and machine learning algorithms. Results show a correlation between window size and forecasting step size for the Support Vector Machines approach but not for the other approaches.

Online financial textual information contains a large amount of investor sentiment, i.e. subjective assessment and discussion with respect to financial instruments. An effective solution to automate the sentiment analysis of such large amounts of online financial texts would be extremely beneficial. This paper presents a natural language processing (NLP) based pre-processing approach both for noise removal from raw online financial texts and for organizing such texts into an enhanced format that is more usable for feature extraction. The proposed approach integrates six NLP processing steps, including a developed syntactic and semantic combined negation handling algorithm, to reduce noise in the online informal text. Three-class sentiment classification is also introduced in each system implementation. Experimental results show that the proposed pre-processing approach outperforms other pre-processing methods. The combined negation handling algorithm is also evaluated against three standard negation handling approaches.

This paper proposes a new locally adaptive boundary evolution algorithm for level set methods (LSM)-based novelty detection. The proposed approach consists of level set function construction, boundary evolution, and evolution termination. It utilises the exterior data points lying outside the decision boundary to effect the segments of the boundary that need to be locally evolved in order to make the boundary better fit the data distribution, so it can evolve boundary locally without requiring knowing explicitly the decision boundary. The experimental results demonstrate that the proposed approach can effectively detect novel events as compared to the reported LSM-based novelty detection method with global boundary evolution scheme and four representative novelty detection methods when there is an exacting error requirement on normal events.

Price manipulation refers to the act of using illegal trading behaviour to manually change an equity price with the aim of making profits. With increasing volumes of trading, price manipulation can be extremely damaging to the proper functioning and integrity of capital markets. Effective approaches for analysing and real-time detection of price manipulation are yet to be developed. This paper proposes a novel approach, called Hidden Markov Model with Abnormal States (HMMAS), which models and detects price manipulation activities. Together with the wavelet decomposition for features extraction and Gaussian Mixture Model for Probability Density Function (PDF) construction, the HMMAS model detects price manipulation and identifies the type of the detected manipulation. Evaluation experiments of the model were conducted on six stock tick data from NASDAQ and London Stock Exchange (LSE). The results showed that the proposed HMMAS model can effectively detect price manipulation patterns.

This paper proposes a locally adaptive level set boundary description (LALSBD) method for novelty detection. The proposed method adjusts the nonlinear boundary directly in the input space and consists of a number of processes including level set function (LSF) construction, local boundary evolution and termination. It employs kernel density estimation (KDE) to construct the LSF and form the initial boundary surrounding the training data. In order to make the boundary better fit the data distribution, a data-driven based local expanding/shrinking evolution method is proposed instead of the global evolution approach reported in our previous level set boundary description (LSBD) method. The proposed LALSBD is compared with LSBD and other four representative novelty detection methods. The experimental results demonstrate that LALSBD can detect novel events more accurately, especially for applications which demand very high classification accuracy for normal events.

This paper introduces a novel neuro-dynamic system for adaptive online clustering using populations of spiking neurons and spike-timing dependent plasticity (STDP). Real-valued data samples are temporally encoded into spike events, used by biological neurons to encode information and communicate with one another, and clusters are represented by spiking neuron populations of varying size. The number of clusters is unknown a priori and clusters are learned in an online fashion where each data sample is provided only once. The coincidence detection capability of spiking neurons is utilized for data clustering and clusters are dynamically formed. The structure of the spiking neural network is constantly adjusted through adding and pruning of neuron populations. Besides, the number of neurons within each population constantly adapts as new data arrives. STDP is employed to adjust the strength of synaptic connections and enhance the selectivity of each population to its corresponding group of data. Preliminary experiments were carried out on synthetic and selected benchmark datasets to evaluate the performance of the proposed system. Promising results were obtained, which indicate the viability of spike-based population coding for online data clustering.

A reliable novelty detector employs a model that encloses the normal dataset tightly. As nonparametric probability density function estimation methods make no assumptions about the probability distribution of a dataset, this paper applies kernel density estimation to construct the initial boundaries surrounding the normal data points. Afterwards, the level set method makes the initial boundaries shrink or expand to better fit the normal data distribution and optimize the boundary surfaces. The proposed method is able to smooth the boundary's evolution automatically while merging or splitting happens. The boundary motion is governed by partial differential equations which formulate the dynamics of the level set method. The proposed novelty detection method is compared with four representative existing methods: support vector data description, nearest neighbours data description, mixture of Gaussian and k-means. The experimental results illustrate that the proposed level set based method presents a comparable performance as mixture of Gaussian, which performs best in terms of false negative and false positive rates.

Recent work by the authors proposed compact low power synapses in hardware, based on the charge-coupling principle, that can be configured to yield a static or dynamic response. The focus of this work is to investigate the training dynamics of these synapses. Empirical models of the Post Synaptic Response (PSP), derived from hardware simulations, were developed and subsequently embedded into the MATLAB environment. A network of these synapses was then used to solve a benchmark problem using a well established training algorithm where the performance metric was convergence time, accuracy and weight range; the Spike Response Model (SRM) was used to implement point neurons. Results are presented and compared with standard synaptic responses.

This study investigates the computing capabilities and potential applications of neural oscillators to grey scale and colour image segmentation, an important task in image understanding and object recognition. A proposed neural system that combines the synergy between neural oscillators and Kohonen self-organising maps (SOM) is presented. Colour image segmentation is achieved through temporal synchronisation of neural oscillators that are mapped to pixels of the same object. Neurons are organised in a two-dimensional grid and are locally connected through excitatory connections and globally connected to a common inhibitor. Self-organising maps form the basis of a colour reduction system whose output is fed to a 2D grid of neural oscillators such as each neuron is mapped to a pixel of the input image. Both chromatic and local spatial features are used. The proposed system is simulated in Matlab and its demonstration on real world colour images shows promising results and the emergence of a new bio-inspired approach for colour image segmentation.

In this work, new structures that implement RSA cryptographic algorithm are presented. These structures are built upon a modified Montgomery modular multiplier, where the operations of multiplication and modular reductions are carried out in parallel rather than interleaved as in the traditional Montgomery multiplier. The global broadcast of data lines is avoided by interleaving two or more encryption/decryption operations onto the same structure, thus making the implementation systolic and scalable. The digit approach has been adopted in This work. This methodology is based on varying the digit size and the level of pipelining of the structures. This parameterised approach presents the designer with an efficient way of choosing the architecture that suits better his/her requirements in terms of speed and area usage, an issue of critical importance to the resources-limited FPGA chips. The results of implementation using FPGA have shown that the proposed RSA structures outperformed those structures built around the traditional Montgomery multiplier in terms of speed, thanks to avoiding global lines broadcast.