Over-the-air computation (OAC) leverages the physical superposition property of wireless multiple access channels (MACs) to compute functions while communication occurs, enabling scalable and low-latency processing in distributed networks. While analog OAC methods suffer from noise sensitivity and hardware constraints, existing digital approaches are often limited in design complexity, which may hinder scalability and fail to exploit spectral efficiency fully. This two-part paper revisits and extends the ChannelComp framework, a general methodology for computing arbitrary finite-valued functions using digital modulation. In Part I, we develop a novel constellation design approach that is aware of the noise distribution and formulates the encoder design as a max-min optimization problem using noise-tailored distance metrics. Our design supports noise models, including Gaussian, Laplace, and heavy-tailed distributions. We further demonstrate that, for heavy-tailed noise, the optimal ChannelComp setup coincides with the solution to the corresponding max-min criterion for the channel noise with heavy-tailed distributions. Numerical experiments confirm that our noise-aware design achieves a substantially lower mean-square error than leading digital OAC methods over noisy MACs. In Part II, we consider a constellation design with a quantization-based sampling scheme to enhance modulation scalability and computational accuracy for large-scale digital OAC.
Edge-cloud collaborative computing (ECCC) has emerged as a pivotal paradigm for addressing the computational demands of modern intelligent applications, integrating cloud resources with edge devices to enable efficient, low-latency processing. Recent advancements in AI, particularly deep learning and large language models (LLMs), have dramatically enhanced the capabilities of these distributed systems, yet introduce significant challenges in model deployment and resource management. In this survey, we comprehensive examine the intersection of distributed intelligence and model optimization within edge-cloud environments, providing a structured tutorial on fundamental architectures, enabling technologies, and emerging applications. Additionally, we systematically analyze model optimization approaches, including compression, adaptation, and neural architecture search, alongside AI-driven resource management strategies that balance performance, energy efficiency, and latency requirements. We further explore critical aspects of privacy protection and security enhancement within ECCC systems and examines practical deployments through diverse applications, spanning autonomous driving, healthcare, and industrial automation. Performance analysis and benchmarking techniques are also thoroughly explored to establish evaluation standards for these complex systems. Furthermore, the review identifies critical research directions including LLMs deployment, 6G integration, neuromorphic computing, and quantum computing, offering a roadmap for addressing persistent challenges in heterogeneity management, real-time processing, and scalability. By bridging theoretical advancements and practical deployments, this survey offers researchers and practitioners a holistic perspective on leveraging AI to optimize distributed computing environments, fostering innovation in next-generation intelligent systems.
The rapid growth of video-text data presents challenges in storage and computation during training. Online learning, which processes streaming data in real-time, offers a promising solution to these issues while also allowing swift adaptations in scenarios demanding real-time responsiveness. One strategy to enhance the efficiency and effectiveness of learning involves identifying and prioritizing data that enhances performance on target downstream tasks. We propose Relevance and Specificity-based online filtering framework (ReSpec) that selects data based on four criteria: (i) modality alignment for clean data, (ii) task relevance for target focused data, (iii) specificity for informative and detailed data, and (iv) efficiency for low-latency processing. Relevance is determined by the probabilistic alignment of incoming data with downstream tasks, while specificity employs the distance to a root embedding representing the least specific data as an efficient proxy for informativeness. By establishing reference points from target task data, ReSpec filters incoming data in real-time, eliminating the need for extensive storage and compute. Evaluating on large-scale datasets WebVid2M and VideoCC3M, ReSpec attains state-of-the-art performance on five zeroshot video retrieval tasks, using as little as 5% of the data while incurring minimal compute. The source code is available at https://github.com/cdjkim/ReSpec.
Event cameras asynchronously output low-latency event streams, promising for state estimation in high-speed motion and challenging lighting conditions. As opposed to frame-based cameras, the motion-dependent nature of event cameras presents persistent challenges in achieving robust event feature detection and matching. In recent years, learning-based approaches have demonstrated superior robustness over traditional handcrafted methods in feature detection and matching, particularly under aggressive motion and HDR scenarios. In this paper, we propose SuperEIO, a novel framework that leverages the learning-based event-only detection and IMU measurements to achieve event-inertial odometry. Our event-only feature detection employs a convolutional neural network under continuous event streams. Moreover, our system adopts the graph neural network to achieve event descriptor matching for loop closure. The proposed system utilizes TensorRT to accelerate the inference speed of deep networks, which ensures low-latency processing and robust real-time operation on resource-limited platforms. Besides, we evaluate our method extensively on multiple public datasets, demonstrating its superior accuracy and robustness compared to other state-of-the-art event-based methods. We have also open-sourced our pipeline to facilitate research in the field: https://github.com/arclab-hku/SuperEIO.
Wearable health devices have a strong demand in real-time biomedical signal processing. However traditional methods often require data transmission to centralized processing unit with substantial computational resources after collecting it from edge devices. Neuromorphic computing is an emerging field that seeks to design specialized hardware for computing systems inspired by the structure, function, and dynamics of the human brain, offering significant advantages in latency and power consumption. This paper explores a novel neuromorphic implementation approach for gesture recognition by extracting spatiotemporal spiking information from surface electromyography (sEMG) data in an event-driven manner. At the same time, the network was designed by implementing a simple-structured and hardware-friendly Physical Reservoir Computing (PRC) framework called Rotating Neuron Reservoir (RNR) within the domain of Spiking neural network (SNN). The spiking RNR (sRNR) is promising to pipeline an innovative solution to compact embedded wearable systems, enabling low-latency, real-time processing directly at the sensor level. The proposed system was validated by an open-access large-scale sEMG database and achieved an average classification accuracy of 74.6\% and 80.3\% using a classical machine learning classifier and a delta learning rule algorithm respectively. While the delta learning rule could be fully spiking and implementable on neuromorphic chips, the proposed gesture recognition system demonstrates the potential for near-sensor low-latency processing.
Large language models (LLMs) have demonstrated remarkable capabilities in handling complex dialogue tasks without requiring use case-specific fine-tuning. However, analyzing live dialogues in real-time necessitates low-latency processing systems, making it impractical to deploy models with billions of parameters due to latency constraints. As a result, practitioners often prefer smaller models with millions of parameters, trained on high-quality, human-annotated datasets. Yet, curating such datasets is both time-consuming and costly. Consequently, there is a growing need to combine the scalability of LLM-generated labels with the precision of human annotations, enabling fine-tuned smaller models to achieve both higher speed and accuracy comparable to larger models. In this paper, we introduce a simple yet effective framework to address this challenge. Our approach is specifically designed for per-utterance classification problems, which encompass tasks such as intent detection, dialogue state tracking, and more. To mitigate the impact of labeling errors from LLMs -- the primary source of inaccuracies in student models -- we propose a noise-reduced preference learning loss. Experimental results demonstrate that our method significantly improves accuracy across utterance-level dialogue tasks, including sentiment detection (over $2\%$), dialogue act classification (over $1.5\%$), etc.
Active vision enables dynamic visual perception, offering an alternative to static feedforward architectures in computer vision, which rely on large datasets and high computational resources. Biological selective attention mechanisms allow agents to focus on salient Regions of Interest (ROIs), reducing computational demand while maintaining real-time responsiveness. Event-based cameras, inspired by the mammalian retina, enhance this capability by capturing asynchronous scene changes enabling efficient low-latency processing. To distinguish moving objects while the event-based camera is in motion the agent requires an object motion segmentation mechanism to accurately detect targets and center them in the visual field (fovea). Integrating event-based sensors with neuromorphic algorithms represents a paradigm shift, using Spiking Neural Networks to parallelize computation and adapt to dynamic environments. This work presents a Spiking Convolutional Neural Network bioinspired attention system for selective attention through object motion sensitivity. The system generates events via fixational eye movements using a Dynamic Vision Sensor integrated into the Speck neuromorphic hardware, mounted on a Pan-Tilt unit, to identify the ROI and saccade toward it. The system, characterized using ideal gratings and benchmarked against the Event Camera Motion Segmentation Dataset, reaches a mean IoU of 82.2% and a mean SSIM of 96% in multi-object motion segmentation. The detection of salient objects reaches 88.8% accuracy in office scenarios and 89.8% in low-light conditions on the Event-Assisted Low-Light Video Object Segmentation Dataset. A real-time demonstrator shows the system's 0.12 s response to dynamic scenes. Its learning-free design ensures robustness across perceptual scenes, making it a reliable foundation for real-time robotic applications serving as a basis for more complex architectures.
Achieving fast and reliable temporal signal encoding is crucial for low-power, always-on systems. While current spike-based encoding algorithms rely on complex networks or precise timing references, simple and robust encoding models can be obtained by leveraging the intrinsic properties of analog hardware substrates. We propose an encoding framework inspired by biological principles that leverages intrinsic neuronal variability to robustly encode continuous stimuli into spatio-temporal patterns, using at most one spike per neuron. The encoder has low model complexity, relying on a shallow network of heterogeneous neurons. It relies on an internal time reference, allowing for continuous processing. Moreover, stimulus parameters can be linearly decoded from the spiking patterns, granting fast information retrieval. Our approach, validated on both analog neuromorphic hardware and simulation, demonstrates high robustness to noise, spike jitter, and reduced heterogeneity. Consistently with biological observations, we observed the spontaneous emergence of patterns with stereotyped spiking order. The proposed encoding scheme facilitates fast, robust and continuous information processing, making it well-suited for low-power, low-latency processing of temporal data on analog neuromorphic substrates.
Invasive cortical brain-machine interfaces (BMIs) can significantly improve the life quality of motor-impaired patients. Nonetheless, externally mounted pedestals pose an infection risk, which calls for fully implanted systems. Such systems, however, must meet strict latency and energy constraints while providing reliable decoding performance. While recurrent spiking neural networks (RSNNs) are ideally suited for ultra-low-power, low-latency processing on neuromorphic hardware, it is unclear whether they meet the above requirements. To address this question, we trained RSNNs to decode finger velocity from cortical spike trains (CSTs) of two macaque monkeys. First, we found that a large RSNN model outperformed existing feedforward spiking neural networks (SNNs) and artificial neural networks (ANNs) in terms of their decoding accuracy. We next developed a tiny RSNN with a smaller memory footprint, low firing rates, and sparse connectivity. Despite its reduced computational requirements, the resulting model performed substantially better than existing SNN and ANN decoders. Our results thus demonstrate that RSNNs offer competitive CST decoding performance under tight resource constraints and are promising candidates for fully implanted ultra-low-power BMIs with the potential to revolutionize patient care.
For Edge AI applications, deploying online learning and adaptation on resource-constrained embedded devices can deal with fast sensor-generated streams of data in changing environments. However, since maintaining low-latency and power-efficient inference is paramount at the Edge, online learning and adaptation on the device should impose minimal additional overhead for inference. With this goal in mind, we explore energy-efficient learning and adaptation on-device for streaming-data Edge AI applications using Spiking Neural Networks (SNNs), which follow the principles of brain-inspired computing, such as high-parallelism, neuron co-located memory and compute, and event-driven processing. We propose EON-1, a brain-inspired processor for near-sensor extreme edge online feature extraction, that integrates a fast online learning and adaptation algorithm. We report results of only 1% energy overhead for learning, by far the lowest overhead when compared to other SoTA solutions, while attaining comparable inference accuracy. Furthermore, we demonstrate that EON-1 is up for the challenge of low-latency processing of HD and UHD streaming video in real-time, with learning enabled.
Machine learning-assisted network security may significantly contribute to securing 5G components. However, machine learning network security inference generally requires tens to hundreds of milliseconds, thereby introducing significant latency in 5G operations. The inference latency can be reduced by deploying the machine learning model to programmable logic in a field programmable gate array at the cost of a small loss in accuracy. In order to quantify this loss, as well as to establish baseline performance inference latency for programmable logic implementations, this work explores an autoencoder and a β-variational autoencoder deployed on two different field programmable gate array evaluation boards and compares accuracy and performance against an NVIDIA A100 graphics processing unit implementation. A publicly available 5G dataset containing 10 types of attacks along with normal traffic is introduced as part of the evaluation.
Identifying optimal join orders (JOs) stands out as a key challenge in database research and engineering. Owing to the large search space, established classical methods rely on approximations and heuristics. Recent efforts have successfully explored reinforcement learning (RL) for JO. Likewise, quantum versions of RL have received considerable scientific attention. Yet, it is an open question if they can achieve sustainable, overall practical advantages with improved quantum processors. In this paper, we present a novel approach that uses quantum reinforcement learning (QRL) for JO based on a hybrid variational quantum ansatz. It is able to handle general bushy join trees instead of resorting to simpler left-deep variants as compared to approaches based on quantum(-inspired) optimisation, yet requires multiple orders of magnitudes fewer qubits, which is a scarce resource even for post-NISQ systems. Despite moderate circuit depth, the ansatz exceeds current NISQ capabilities, which requires an evaluation by numerical simulations. While QRL may not significantly outperform classical approaches in solving the JO problem with respect to result quality (albeit we see parity), we find a drastic reduction in required trainable parameters. This benefits practically relevant aspects ranging from shorter training times compared to classical RL, less involved classical optimisation passes, or better use of available training data, and fits data-stream and low-latency processing scenarios. Our comprehensive evaluation and careful discussion delivers a balanced perspective on possible practical quantum advantage, provides insights for future systemic approaches, and allows for quantitatively assessing trade-offs of quantum approaches for one of the most crucial problems of database management systems.
This paper introduces a novel low-latency online beamforming (BF) algorithm, named Modified Parametric Multichannel Wiener Filter (Mod-PMWF), for enhancing speech mixtures with unknown and varying number of speakers. Although conventional BFs such as linearly constrained minimum variance BF (LCMV BF) can enhance a speech mixture, they typically require such attributes of the speech mixture as the number of speakers and the acoustic transfer functions (ATFs) from the speakers to the microphones. When the mixture attributes are unavailable, estimating them by low-latency processing is challenging, hindering the application of the BFs to the problem. In this paper, we overcome this problem by modifying a conventional Parametric Multichannel Wiener Filter (PMWF). The proposed Mod-PMWF can adaptively form a directivity pattern that enhances all the speakers in the mixture without explicitly estimating these attributes. Our experiments will show the proposed BF's effectiveness in interference reduction ratios and subjective listening tests.
The concept of neuromuscular activity recognition using instantaneous high-density surface electromyography (HD-sEMG) images opens up new avenues for the development of more fluid and natural muscle-computer interfaces. However, the existing approaches employed a very large deep convolutional neural network (ConvNet) architecture and complex training schemes for HD-sEMG image recognition, which requires the network architecture to be pre-trained on a very large-scale labeled training dataset, as a result, it makes computationally very expensive. To overcome this problem, we propose S-ConvNet and All-ConvNet models, a simple yet efficient framework for learning instantaneous HD-sEMG images from scratch for neuromuscular activity recognition. Without using any pre-trained models, our proposed S-ConvNet and All-ConvNet demonstrate very competitive recognition accuracy to the more complex state of the art for neuromuscular activity recognition based on instantaneous HD-sEMG images, while using a ~ 12 x smaller dataset and reducing learning parameters to a large extent. The experimental results proved that the S-ConvNet and All-ConvNet are highly effective for learning discriminative features for instantaneous HD-sEMG image recognition especially in the data and high-end resource constrained scenarios.
In many experiment-driven scientific domains, such as high-energy physics, material science, and cosmology, high data rate experiments impose hard constraints on data acquisition systems: collected data must either be indiscriminately stored for post-processing and analysis, thereby necessitating large storage capacity, or accurately filtered in real-time, thereby necessitating low-latency processing. Deep neural networks, effective in other filtering tasks, have not been widely employed in such data acquisition systems, due to design and deployment difficulties. We present an open source, lightweight, compiler framework, without any proprietary dependencies, OpenHLS, based on high-level synthesis techniques, for translating high-level representations of deep neural networks to low-level representations, suitable for deployment to near-sensor devices such as field-programmable gate arrays. We evaluate OpenHLS on various workloads and present a case-study implementation of a deep neural network for Bragg peak detection in the context of high-energy diffraction microscopy. We show OpenHLS is able to produce an implementation of the network with a throughput 4.8 $\mu$s/sample, which is approximately a 4$\times$ improvement over the existing implementation
Streaming ASR with strict latency constraints is required in many speech recognition applications. In order to achieve the required latency, streaming ASR models sacrifice accuracy compared to non-streaming ASR models due to lack of future input context. Previous research has shown that streaming and non-streaming ASR for RNN Transducers can be unified by cascading causal and non-causal encoders. This work improves upon this cascaded encoders framework by leveraging two streaming non-causal encoders with variable input context sizes that can produce outputs at different audio intervals (e.g. fast and slow). We propose a novel parallel time-synchronous beam search algorithm for transducers that decodes from fast-slow encoders, where the slow encoder corrects the mistakes generated from the fast encoder. The proposed algorithm, achieves up to 20% WER reduction with a slight increase in token emission delays on the public Librispeech dataset and in-house datasets. We also explore techniques to reduce the computation by distributing processing between the fast and slow encoders. Lastly, we explore sharing the parameters in the fast encoder to reduce the memory footprint. This enables low latency processing on edge devices with low computation cost and a low memory footprint.
This paper presents a novel high-fidelity and low-latency universal neural vocoder framework based on multiband WaveRNN with data-driven linear prediction for discrete waveform modeling (MWDLP). MWDLP employs a coarse-fine bit WaveRNN architecture for 10-bit mu-law waveform modeling. A sparse gated recurrent unit with a relatively large size of hidden units is utilized, while the multiband modeling is deployed to achieve real-time low-latency usage. A novel technique for data-driven linear prediction (LP) with discrete waveform modeling is proposed, where the LP coefficients are estimated in a data-driven manner. Moreover, a novel loss function using short-time Fourier transform (STFT) for discrete waveform modeling with Gumbel approximation is also proposed. The experimental results demonstrate that the proposed MWDLP framework generates high-fidelity synthetic speech for seen and unseen speakers and/or language on 300 speakers training data including clean and noisy/reverberant conditions, where the number of training utterances is limited to 60 per speaker, while allowing for real-time low-latency processing using a single core of $\sim\!$ 2.1--2.7 GHz CPU with $\sim\!$ 0.57--0.64 real-time factor including input/output and feature extraction.
In studies of the visual system as well as in computer vision, the focus is often on contrast edges. However, the primate visual system contains a large number of cells that are insensitive to spatial contrast and, instead, respond to uniform homogeneous illumination of their visual field. The purpose of this information remains unclear. Here, we propose a mechanism that detects feature homogeneity in visual areas, based on latency coding and spike time coincidence, in a purely feed-forward and therefore rapid manner. We demonstrate how homogeneity information can interact with information on contrast edges to potentially support rapid image segmentation. Furthermore, we analyze how neuronal crosstalk (noise) affects the mechanism's performance. We show that the detrimental effects of crosstalk can be partly mitigated through delayed feed-forward inhibition that shapes bi-phasic post-synaptic events. The delay of the feed-forward inhibition allows effectively controlling the size of the temporal integration window and, thereby, the coincidence threshold. The proposed model is based on single-spike latency codes in a purely feed-forward architecture that supports low-latency processing, making it an attractive scheme of computation in spiking neuronal networks where rapid responses and low spike counts are desired.
Neuromorphic systems that learn and predict from streaming inputs hold significant promise in pervasive edge computing and its applications. In this paper, a neuromorphic system that processes spatio-temporal information on the edge is proposed. Algorithmically, the system is based on hierarchical temporal memory that inherently offers online learning, resiliency, and fault tolerance. Architecturally, it is a full custom mixed-signal design with an underlying digital communication scheme and analog computational modules. Therefore, the proposed system features reconfigurability, real-time processing, low power consumption, and low-latency processing. The proposed architecture is benchmarked to predict on real-world streaming data. The network's mean absolute percentage error on the mixed-signal system is 1.129X lower compared to its baseline algorithm model. This reduction can be attributed to device non-idealities and probabilistic formation of synaptic connections. We demonstrate that the combined effect of Hebbian learning and network sparsity also plays a major role in extending the overall network lifespan. We also illustrate that the system offers 3.46X reduction in latency and 77.02X reduction in power consumption when compared to a custom CMOS digital design implemented at the same technology node. By employing specific low power techniques, such as clock gating, we observe 161.37X reduction in power consumption.
In this work, we investigate if the learned encoder of the end-to-end convolutional time domain audio separation network (Conv-TasNet) is the key to its recent success, or if the encoder can just as well be replaced by a deterministic hand-crafted filterbank. Motivated by the resemblance of the trained encoder of Conv-TasNet to auditory filterbanks, we propose to employ a deterministic gammatone filterbank. In contrast to a common gammatone filterbank, our filters are restricted to 2 ms length to allow for low-latency processing. Inspired by the encoder learned by Conv-TasNet, in addition to the logarithmically spaced filters, the proposed filterbank holds multiple gammatone filters at the same center frequency with varying phase shifts. We show that replacing the learned encoder with our proposed multi-phase gammatone filterbank (MP-GTF) even leads to a scale-invariant source-to-noise ratio (SI-SNR) improvement of 0.7 dB. Furthermore, in contrast to using the learned encoder we show that the number of filters can be reduced from 512 to 128 without loss of performance.