Nikil Dutt

Mingoo Ji, Saehanseul Yi, Changjin Koo et al.

When executing a deep neural network (DNN), its model parameters are loaded into GPU memory before execution, incurring a significant GPU memory burden. There are studies that reduce GPU memory usage by exploiting CPU memory as a swap device. However, this approach is not applicable in most embedded systems with integrated GPUs where CPU and GPU share a common memory. In this regard, we present Demand Layering, which employs a fast solid-state drive (SSD) as a co-running partner of a GPU and exploits the layer-by-layer execution of DNNs. In our approach, a DNN is loaded and executed in a layer-by-layer manner, minimizing the memory usage to the order of a single layer. Also, we developed a pipeline architecture that hides most additional delays caused by the interleaved parameter loadings alongside layer executions. Our implementation shows a 96.5% memory reduction with just 14.8% delay overhead on average for representative DNNs. Furthermore, by exploiting the memory-delay tradeoff, near-zero delay overhead (under 1 ms) can be achieved with a slightly increased memory usage (still an 88.4% reduction), showing the great potential of Demand Layering.

Anil Kanduri, Sina Shahhosseini, Emad Kasaeyan Naeini et al.

Smart eHealth applications deliver personalized and preventive digital healthcare services to clients through remote sensing, continuous monitoring, and data analytics. Smart eHealth applications sense input data from multiple modalities, transmit the data to edge and/or cloud nodes, and process the data with compute intensive machine learning (ML) algorithms. Run-time variations with continuous stream of noisy input data, unreliable network connection, computational requirements of ML algorithms, and choice of compute placement among sensor-edge-cloud layers affect the efficiency of ML-driven eHealth applications. In this chapter, we present edge-centric techniques for optimized compute placement, exploration of accuracy-performance trade-offs, and cross-layered sense-compute co-optimization for ML-driven eHealth applications. We demonstrate the practical use cases of smart eHealth applications in everyday settings, through a sensor-edge-cloud framework for an objective pain assessment case study.

Ali Tazarv, Sina Labbaf, Amir Rahmani et al.

Most existing sensor-based monitoring frameworks presume that a large available labeled dataset is processed to train accurate detection models. However, in settings where personalization is necessary at deployment time to fine-tune the model, a person-specific dataset needs to be collected online by interacting with the users. Optimizing the collection of labels in such phase is instrumental to impose a tolerable burden on the users while maximizing personal improvement. In this paper, we consider a fine-grain stress detection problem based on wearable sensors targeting everyday settings, and propose a novel context-aware active learning strategy capable of jointly maximizing the meaningfulness of the signal samples we request the user to label and the response rate. We develop a multilayered sensor-edge-cloud platform to periodically capture physiological signals and process them in real-time, as well as to collect labels and retrain the detection model. We collect a large dataset and show that the context-aware active learning technique we propose achieves a desirable detection performance using 88\% and 32\% fewer queries from users compared to a randomized strategy and a traditional active learning strategy, respectively.

Sina Shahhosseini, Yang Ni, Hamidreza Alikhani et al.

Health monitoring applications increasingly rely on machine learning techniques to learn end-user physiological and behavioral patterns in everyday settings. Considering the significant role of wearable devices in monitoring human body parameters, on-device learning can be utilized to build personalized models for behavioral and physiological patterns, and provide data privacy for users at the same time. However, resource constraints on most of these wearable devices prevent the ability to perform online learning on them. To address this issue, it is required to rethink the machine learning models from the algorithmic perspective to be suitable to run on wearable devices. Hyperdimensional computing (HDC) offers a well-suited on-device learning solution for resource-constrained devices and provides support for privacy-preserving personalization. Our HDC-based method offers flexibility, high efficiency, resilience, and performance while enabling on-device personalization and privacy protection. We evaluate the efficacy of our approach using three case studies and show that our system improves the energy efficiency of training by up to $45.8\times$ compared with the state-of-the-art Deep Neural Network (DNN) algorithms while offering a comparable accuracy.

Zexin Li, Nikil Dutt, Cong Liu

Online continual learning (OCL) enables real-time adaptation to new data, making it crucial for dynamic robotic applications. However, its practical deployment is hindered by memory constraints in resource-limited systems, which affect key trade-offs in training latency, plasticity, and stability. Unlike offline parameter tuning, which cannot account for the dynamic shift in memory pressure and workload complexity as OCL progresses, an online and self-adaptive approach is essential for robust on-device deployment. This paper proposes Orion, a holistic framework designed to co-optimize training latency, plasticity, and stability of state-of-the-art OCL models under strict memory constraints, enabling feasible on-device deployment. At its core, Orion leverages URGE, a unified runtime indicator grounded in the ``Buckets effect'' principle that system performance is bounded by its scarcest resource, to dynamically reallocate memory across OCL components by jointly coordinating batch processing, replay buffers, and optimization strategies at both the OS and application level. Furthermore, Orion introduces system-level data prefetching techniques to maximize efficiency. A system prototype of Orion has been implemented using the widely adopted \texttt{Avalanche-lib} and thoroughly evaluated across a diverse range of OCL algorithms, benchmarks, and hardware platforms commonly used in autonomous robotic applications. To further demonstrate its practical utility, Orion is integrated into a realistic autonomous navigational robot powered by OCL. The results show that Orion achieves significant training speedups while maintaining balanced performance and effectively adapting to various scenarios, all with minimal runtime, memory, and energy overhead, making Orion a practical solution for on-device continual learning.

Zhongqi Yang, Iman Azimi, Salar Jafarlou et al.

The adverse effects of loneliness on both physical and mental well-being are profound. Although previous research has utilized mobile sensing techniques to detect mental health issues, few studies have utilized state-of-the-art wearable devices to forecast loneliness and estimate the physiological manifestations of loneliness and its predictive nature. The primary objective of this study is to examine the feasibility of forecasting loneliness by employing wearable devices, such as smart rings and watches, to monitor early physiological indicators of loneliness. Furthermore, smartphones are employed to capture initial behavioral signs of loneliness. To accomplish this, we employed personalized machine learning techniques, leveraging a comprehensive dataset comprising physiological and behavioral information obtained during our study involving the monitoring of college students. Through the development of personalized models, we achieved a notable accuracy of 0.82 and an F-1 score of 0.82 in forecasting loneliness levels seven days in advance. Additionally, the application of Shapley values facilitated model explainability. The wealth of data provided by this study, coupled with the forecasting methodology employed, possesses the potential to augment interventions and facilitate the early identification of loneliness within populations at risk.

Seyed Amir Hossein Aqajari, Ziyu Wang, Ali Tazarv et al.

In today's fast-paced world, accurately monitoring stress levels is crucial. Sensor-based stress monitoring systems often need large datasets for training effective models. However, individual-specific models are necessary for personalized and interactive scenarios. Traditional methods like Ecological Momentary Assessments (EMAs) assess stress but struggle with efficient data collection without burdening users. The challenge is to timely send EMAs, especially during stress, balancing monitoring efficiency and user convenience. This paper introduces a novel context-aware active reinforcement learning (RL) algorithm for enhanced stress detection using Photoplethysmography (PPG) data from smartwatches and contextual data from smartphones. Our approach dynamically selects optimal times for deploying EMAs, utilizing the user's immediate context to maximize label accuracy and minimize intrusiveness. Initially, the study was executed in an offline environment to refine the label collection process, aiming to increase accuracy while reducing user burden. Later, we integrated a real-time label collection mechanism, transitioning to an online methodology. This shift resulted in an 11% improvement in stress detection efficiency. Incorporating contextual data improved model accuracy by 4%. Personalization studies indicated a 10% enhancement in AUC-ROC scores, demonstrating better stress level differentiation. This research marks a significant move towards personalized, context-driven real-time stress monitoring methods.

Manoj Vishwanath, Steven Cao, Nikil Dutt et al.

While analytics of sleep electroencephalography (EEG) holds certain advantages over other methods in clinical applications, high variability across subjects poses a significant challenge when it comes to deploying machine learning models for classification tasks in the real world. In such instances, machine learning models that exhibit exceptional performance on a specific dataset may not necessarily demonstrate similar proficiency when applied to a distinct dataset for the same task. The scarcity of high-quality biomedical data further compounds this challenge, making it difficult to evaluate the model's generality comprehensively. In this paper, we introduce Transfer Euclidean Alignment - a transfer learning technique to tackle the problem of the dearth of human biomedical data for training deep learning models. We tested the robustness of this transfer learning technique on various rule-based classical machine learning models as well as the EEGNet-based deep learning model by evaluating on different datasets, including human and mouse data in a binary classification task of detecting individuals with versus without traumatic brain injury (TBI). By demonstrating notable improvements with an average increase of 14.42% for intraspecies datasets and 5.53% for interspecies datasets, our findings underscore the importance of the use of transfer learning to improve the performance of machine learning and deep learning models when using diverse datasets for training.

Nazmul Takbir, Hamidreza Alikhani, Nikil Dutt et al.

Large Language Model (LLM) serving is increasingly constrained by the growing size of the key-value (KV) cache, which scales with both context length and generation length. Prior work shows that attention is dominated by a small subset of critical tokens, yet existing systems struggle to exploit this efficiently without degrading accuracy, especially in long generation. We make a key observation: the temporal stability of these critical tokens varies significantly across KV heads: some heads consistently focus on the same tokens, while others shift frequently. Building on this insight, we introduce FlexiCache, a hierarchical KV-cache management system that leverages the temporal stability of KV heads to reduce GPU memory usage and computation overhead, while preserving model accuracy. FlexiCache classifies KV heads as stable or unstable: it retains all KV-cache pages from unstable heads in GPU memory, whereas for stable heads, it keeps only the top-K pages on the GPU and offloads the rest to host memory. By exploiting temporal stability, FlexiCache performs periodic reranking for stable heads to fetch newly promoted top pages. Implemented atop vLLM, FlexiCache reduces GPU memory footprint for long-context requests by up to 70%, improves offline serving throughput by 1.38-1.55x, and lowers online token latency by 1.6-2.1x, all while maintaining accuracy in long-context, long-generation scenarios.

Danny Abraham, Nikhil Kamalkumar Advani, Arun Das et al.

Accurate 3D lane segment detection and topology reasoning are critical for structured online map construction in autonomous driving. Recent transformer-based approaches formulate this task as query-based set prediction, yet largely inherit decoder designs originally developed for compact object detection. However, lane segments are continuous polylines embedded in directed graphs, and generic query initialization and unconstrained refinement do not explicitly encode this geometric and relational structure. We propose GeoReFormer (Geometry-aware Refinement Transformer), a unified query-based architecture that embeds geometry- and topology-aware inductive biases directly within the transformer decoder. GeoReFormer introduces data-driven geometric priors for structured query initialization, bounded coordinate-space refinement for stable polyline deformation, and per-query gated topology propagation to selectively integrate relational context. On the OpenLane-V2 benchmark, GeoReFormer achieves state-of-the-art performance with 34.5% mAP while improving topology consistency over strong transformer baselines, demonstrating the utility of explicit geometric and relational structure encoding.

Yutong Song, Shiva Shrestha, Chenhan Lyu et al.

Spoken question-answering (SQA) systems relying on automatic speech recognition (ASR) often struggle with accurately recognizing medical terminology. To this end, we propose MedSpeak, a novel knowledge graph-aided ASR error correction framework that refines noisy transcripts and improves downstream answer prediction by leveraging both semantic relationships and phonetic information encoded in a medical knowledge graph, together with the reasoning power of LLMs. Comprehensive experimental results on benchmarks demonstrate that MedSpeak significantly improves the accuracy of medical term recognition and overall medical SQA performance, establishing MedSpeak as a state-of-the-art solution for medical SQA. The code is available at https://github.com/RainieLLM/MedSpeak.

Cheonkam Jeong, Jessica Liao, Audrey Lu et al.

We present DementiaBank-Emotion, the first multi-rater emotion annotation corpus for Alzheimer's disease (AD) speech. Annotating 1,492 utterances from 108 speakers for Ekman's six basic emotions and neutral, we find that AD patients express significantly more non-neutral emotions (16.9%) than healthy controls (5.7%; p < .001). Exploratory acoustic analysis suggests a possible dissociation: control speakers showed substantial F0 modulation for sadness (Delta = -3.45 semitones from baseline), whereas AD speakers showed minimal change (Delta = +0.11 semitones; interaction p = .023), though this finding is based on limited samples (sadness: n=5 control, n=15 AD) and requires replication. Within AD speech, loudness differentiates emotion categories, indicating partially preserved emotion-prosody mappings. We release the corpus, annotation guidelines, and calibration workshop materials to support research on emotion recognition in clinical populations.

Seyed Amir Hossein Aqajari, Emad Kasaeyan Naeini, Milad Asgari Mehrabadi et al.

The stress detection problem is receiving great attention in related research communities. This is due to its essential part in behavioral studies for many serious health problems and physical illnesses. There are different methods and algorithms for stress detection using different physiological signals. Previous studies have already shown that Galvanic Skin Response (GSR), also known as Electrodermal Activity (EDA), is one of the leading indicators for stress. However, the GSR signal itself is not trivial to analyze. Different features are extracted from GSR signals to detect stress in people like the number of peaks, max peak amplitude, etc. In this paper, we are proposing an open-source tool for GSR analysis, which uses deep learning algorithms alongside statistical algorithms to extract GSR features for stress detection. Then we use different machine learning algorithms and Wearable Stress and Affect Detection (WESAD) dataset to evaluate our results. The results show that we are capable of detecting stress with the accuracy of 92 percent using 10-fold cross-validation and using the features extracted from our tool.

Yutong Song, Jiang Wu, Weijia Zhang et al.

Adapting large language models to individual users remains challenging due to the tension between fine-grained personalization and scalable deployment. We present CARD, a hierarchical framework that achieves effective personalization through progressive refinement. CARD first clusters users according to shared stylistic patterns and learns cluster-specific LoRA adapters, enabling robust generalization and strong low-resource performance. To capture individual differences within each cluster, we propose an implicit preference learning mechanism that contrasts user-authored text with cluster-level generations, allowing the model to infer user-specific style preferences without manual annotation. At inference time, CARD injects personalization exclusively at decoding via lightweight user preference vectors and low-rank logit corrections, while keeping the base model frozen. Experiments on the LaMP and LongLaMP benchmarks show that CARD achieves competitive or superior generation quality compared to state-of-the-art baselines, while significantly improving efficiency and scalability for practical personalized text generation.

Zhongqi Yang, Yuning Wang, Ken S. Yamashita et al.

Emotional states, as indicators of affect, are pivotal to overall health, making their accurate prediction before onset crucial. Current studies are primarily centered on immediate short-term affect detection using data from wearable and mobile devices. These studies typically focus on objective sensory measures, often neglecting other forms of self-reported information like diaries and notes. In this paper, we propose a multimodal deep learning model for affect status forecasting. This model combines a transformer encoder with a pre-trained language model, facilitating the integrated analysis of objective metrics and self-reported diaries. To validate our model, we conduct a longitudinal study, enrolling college students and monitoring them over a year, to collect an extensive dataset including physiological, environmental, sleep, metabolic, and physical activity parameters, alongside open-ended textual diaries provided by the participants. Our results demonstrate that the proposed model achieves predictive accuracy of 82.50% for positive affect and 82.76% for negative affect, a full week in advance. The effectiveness of our model is further elevated by its explainability.

Seyed Amir Hossein Aqajari, Sina Labbaf, Phuc Hoang Tran et al.

Daily monitoring of stress is a critical component of maintaining optimal physical and mental health. Physiological signals and contextual information have recently emerged as promising indicators for detecting instances of heightened stress. Nonetheless, developing a real-time monitoring system that utilizes both physiological and contextual data to anticipate stress levels in everyday settings while also gathering stress labels from participants represents a significant challenge. We present a monitoring system that objectively tracks daily stress levels by utilizing both physiological and contextual data in a daily-life environment. Additionally, we have integrated a smart labeling approach to optimize the ecological momentary assessment (EMA) collection, which is required for building machine learning models for stress detection. We propose a three-tier Internet-of-Things-based system architecture to address the challenges. We utilized a cross-validation technique to accurately estimate the performance of our stress models. We achieved the F1-score of 70\% with a Random Forest classifier using both PPG and contextual data, which is considered an acceptable score in models built for everyday settings. Whereas using PPG data alone, the highest F1-score achieved is approximately 56\%, emphasizing the significance of incorporating both PPG and contextual data in stress detection tasks.

Yutong Song, Chenhan Lyu, Pengfei Zhang et al.

Mild-stage dementia patients primarily experience two critical symptoms: severe memory loss and emotional instability. To address these challenges, we propose DEMENTIA-PLAN, an innovative retrieval-augmented generation framework that leverages large language models to enhance conversational support. Our model employs a multiple knowledge graph architecture, integrating various dimensional knowledge representations including daily routine graphs and life memory graphs. Through this multi-graph architecture, DEMENTIA-PLAN comprehensively addresses both immediate care needs and facilitates deeper emotional resonance through personal memories, helping stabilize patient mood while providing reliable memory support. Our notable innovation is the self-reflection planning agent, which systematically coordinates knowledge retrieval and semantic integration across multiple knowledge graphs, while scoring retrieved content from daily routine and life memory graphs to dynamically adjust their retrieval weights for optimized response generation. DEMENTIA-PLAN represents a significant advancement in the clinical application of large language models for dementia care, bridging the gap between AI tools and caregivers interventions.

Hyunwoo Oh, KyungIn Nam, Rajat Bhattacharjya et al.

Recent advances in LLMs have outpaced the computational and memory capacities of edge platforms that primarily employ CPUs, thereby challenging efficient and scalable deployment. While ternary quantization enables significant resource savings, existing CPU solutions rely heavily on memory-based lookup tables (LUTs) which limit scalability, and FPGA or GPU accelerators remain impractical for edge use. This paper presents T-SAR, the first framework to achieve scalable ternary LLM inference on CPUs by repurposing the SIMD register file for dynamic, in-register LUT generation with minimal hardware modifications. T-SAR eliminates memory bottlenecks and maximizes data-level parallelism, delivering 5.6-24.5x and 1.1-86.2x improvements in GEMM latency and GEMV throughput, respectively, with only 3.2% power and 1.4% area overheads in SIMD units. T-SAR achieves up to 2.5-4.9x the energy efficiency of an NVIDIA Jetson AGX Orin, establishing a practical approach for efficient LLM inference on edge platforms.

Rajat Bhattacharjya, Sing-Yao Wu, Hyunwoo Oh et al.

Unmanned Aerial Vehicles (UAVs) in disaster response require complex, queryable intelligence that on-board CNNs cannot provide. While Vision-Language Models (VLMs) offer this semantic reasoning, their high resource demands make on-device deployment infeasible, and naive cloud offloading fails under the low-bandwidth networks common in disaster zones. We present AVERY, a framework that enables VLM deployment through adaptive split computing. We advance the split computing paradigm beyond traditional depth-wise partitioning by introducing a functional, cognitive-inspired dual-stream split that separates the VLM into a high-frequency, low-resolution "context stream" for real-time awareness and a low-frequency, high-fidelity "insight stream" for deep analysis. A lightweight, self-aware on-board controller manages this architecture, monitoring network conditions and operator intent to dynamically select from pre-trained compression models, navigating the fundamental accuracy-throughput trade-off. Evaluated using the VLM LISA-7B across an edge-cloud scenario under fluctuating network conditions, AVERY consistently outperforms static configurations, achieving 11.2% higher accuracy than raw image compression and 93.98% lower energy consumption compared to full-edge execution, thereby enhancing mission efficiency and enabling real-time, queryable intelligence on resource-constrained platforms in dynamic environments.

Rajat Bhattacharjya, Woohyeok Park, Arnab Sarkar et al.

Direction of Arrival (DoA) estimation techniques face a critical trade-off, as classical methods often lack accuracy in challenging, low signal-to-noise ratio (SNR) conditions, while modern deep learning approaches are too energy-intensive and opaque for resource-constrained, safety-critical systems. We introduce HYPERDOA, a novel estimator leveraging Hyperdimensional Computing (HDC). The framework introduces two distinct feature extraction strategies -- Mean Spatial-Lag Autocorrelation and Spatial Smoothing -- for its HDC pipeline, and then reframes DoA estimation as a pattern recognition problem. This approach leverages HDC's inherent robustness to noise and its transparent algebraic operations to bypass the expensive matrix decompositions and ``black-box'' nature of classical and deep learning methods, respectively. Our evaluation demonstrates that HYPERDOA achieves ~35.39% higher accuracy than state-of-the-art methods in low-SNR, coherent-source scenarios. Crucially, it also consumes ~93% less energy than competing neural baselines on an embedded NVIDIA Jetson Xavier NX platform. This dual advantage in accuracy and efficiency establishes HYPERDOA as a robust and viable solution for mission-critical applications on edge devices.

Sina Shahhosseini, Tianyi Hu, Dongjoo Seo et al.

Deep-learning-based intelligent services have become prevalent in cyber-physical applications including smart cities and health-care. Collaborative end-edge-cloud computing for deep learning provides a range of performance and efficiency that can address application requirements through computation offloading. The decision to offload computation is a communication-computation co-optimization problem that varies with both system parameters (e.g., network condition) and workload characteristics (e.g., inputs). Identifying optimal orchestration considering the cross-layer opportunities and requirements in the face of varying system dynamics is a challenging multi-dimensional problem. While Reinforcement Learning (RL) approaches have been proposed earlier, they suffer from a large number of trial-and-errors during the learning process resulting in excessive time and resource consumption. We present a Hybrid Learning orchestration framework that reduces the number of interactions with the system environment by combining model-based and model-free reinforcement learning. Our Deep Learning inference orchestration strategy employs reinforcement learning to find the optimal orchestration policy. Furthermore, we deploy Hybrid Learning (HL) to accelerate the RL learning process and reduce the number of direct samplings. We demonstrate efficacy of our HL strategy through experimental comparison with state-of-the-art RL-based inference orchestration, demonstrating that our HL strategy accelerates the learning process by up to 166.6x.

Sina Shahhosseini, Dongjoo Seo, Anil Kanduri et al.

Deep-learning-based intelligent services have become prevalent in cyber-physical applications including smart cities and health-care. Deploying deep-learning-based intelligence near the end-user enhances privacy protection, responsiveness, and reliability. Resource-constrained end-devices must be carefully managed in order to meet the latency and energy requirements of computationally-intensive deep learning services. Collaborative end-edge-cloud computing for deep learning provides a range of performance and efficiency that can address application requirements through computation offloading. The decision to offload computation is a communication-computation co-optimization problem that varies with both system parameters (e.g., network condition) and workload characteristics (e.g., inputs). On the other hand, deep learning model optimization provides another source of tradeoff between latency and model accuracy. An end-to-end decision-making solution that considers such computation-communication problem is required to synergistically find the optimal offloading policy and model for deep learning services. To this end, we propose a reinforcement-learning-based computation offloading solution that learns optimal offloading policy considering deep learning model selection techniques to minimize response time while providing sufficient accuracy. We demonstrate the effectiveness of our solution for edge devices in an end-edge-cloud system and evaluate with a real-setup implementation using multiple AWS and ARM core configurations. Our solution provides 35% speedup in the average response time compared to the state-of-the-art with less than 0.9% accuracy reduction, demonstrating the promise of our online learning framework for orchestrating DL inference in end-edge-cloud systems.

Salar Jafarlou, Jocelyn Lai, Zahra Mousavi et al.

Monitoring and understanding affective states are important aspects of healthy functioning and treatment of mood-based disorders. Recent advancements of ubiquitous wearable technologies have increased the reliability of such tools in detecting and accurately estimating mental states (e.g., mood, stress, etc.), offering comprehensive and continuous monitoring of individuals over time. Previous attempts to model an individual's mental state were limited to subjective approaches or the inclusion of only a few modalities (i.e., phone, watch). Thus, the goal of our study was to investigate the capacity to more accurately predict affect through a fully automatic and objective approach using multiple commercial devices. Longitudinal physiological data and daily assessments of emotions were collected from a sample of college students using smart wearables and phones for over a year. Results showed that our model was able to predict next-day affect with accuracy comparable to state of the art methods.

Milad Asgari Mehrabadi, Seyed Amir Hossein Aqajari, Amir Hosein Afandizadeh Zargari et al.

Continuous monitoring of blood pressure (BP)can help individuals manage their chronic diseases such as hypertension, requiring non-invasive measurement methods in free-living conditions. Recent approaches fuse Photoplethysmograph (PPG) and electrocardiographic (ECG) signals using different machine and deep learning approaches to non-invasively estimate BP; however, they fail to reconstruct the complete signal, leading to less accurate models. In this paper, we propose a cycle generative adversarial network (CycleGAN) based approach to extract a BP signal known as ambulatory blood pressure (ABP) from a clean PPG signal. Our approach uses a cycle generative adversarial network that extends theGAN architecture for domain translation, and outperforms state-of-the-art approaches by up to 2x in BP estimation.

Emad Kasaeyan Naeini, Sina Shahhosseini, Anil Kanduri et al.

eHealth systems deliver critical digital healthcare and wellness services for users by continuously monitoring physiological and contextual data. eHealth applications use multi-modal machine learning kernels to analyze data from different sensor modalities and automate decision-making. Noisy inputs and motion artifacts during sensory data acquisition affect the i) prediction accuracy and resilience of eHealth services and ii) energy efficiency in processing garbage data. Monitoring raw sensory inputs to identify and drop data and features from noisy modalities can improve prediction accuracy and energy efficiency. We propose a closed-loop monitoring and control framework for multi-modal eHealth applications, AMSER, that can mitigate garbage-in garbage-out by i) monitoring input modalities, ii) analyzing raw input to selectively drop noisy data and features, and iii) choosing appropriate machine learning models that fit the configured data and feature vector - to improve prediction accuracy and energy efficiency. We evaluate our AMSER approach using multi-modal eHealth applications of pain assessment and stress monitoring over different levels and types of noisy components incurred via different sensor modalities. Our approach achieves up to 22\% improvement in prediction accuracy and 5.6$\times$ energy consumption reduction in the sensing phase against the state-of-the-art multi-modal monitoring application.

Ali Tazarv, Sina Labbaf, Stephanie M. Reich et al.

Since stress contributes to a broad range of mental and physical health problems, the objective assessment of stress is essential for behavioral and physiological studies. Although several studies have evaluated stress levels in controlled settings, objective stress assessment in everyday settings is still largely under-explored due to challenges arising from confounding contextual factors and limited adherence for self-reports. In this paper, we explore the objective prediction of stress levels in everyday settings based on heart rate (HR) and heart rate variability (HRV) captured via low-cost and easy-to-wear photoplethysmography (PPG) sensors that are widely available on newer smart wearable devices. We present a layered system architecture for personalized stress monitoring that supports a tunable collection of data samples for labeling, and present a method for selecting informative samples from the stream of real-time data for labeling. We captured the stress levels of fourteen volunteers through self-reported questionnaires over periods of between 1-3 months, and explored binary stress detection based on HR and HRV using Machine Learning Methods. We observe promising preliminary results given that the dataset is collected in the challenging environments of everyday settings. The binary stress detector is fairly accurate and can detect stressful vs non-stressful samples with a macro-F1 score of up to \%76. Our study lays the groundwork for more sophisticated labeling strategies that generate context-aware, personalized models that will empower health professionals to provide personalized interventions.

Saehanseul Yi, Tae-Wook Kim, Jong-Chan Kim et al.

The increasing computing demands of autonomous driving applications make energy optimizations critical for reducing battery capacity and vehicle weight. Current energy optimization methods typically target traditional real-time systems with static deadlines, resulting in conservative energy savings that are unable to exploit additional energy optimizations due to dynamic deadlines arising from the vehicle's change in velocity and driving context. We present an adaptive system optimization and reconfiguration approach that dynamically adapts the scheduling parameters and processor speeds to satisfy dynamic deadlines while consuming as little energy as possible. Our experimental results with an autonomous driving task set from Bosch and real-world driving data show energy reductions up to 46.4% on average in typical dynamic driving scenarios compared with traditional static energy optimization methods, demonstrating great potential for dynamic energy optimization gains by exploiting dynamic deadlines.

Adarsha Balaji, Shihao Song, Twisha Titirsha et al.

Recently, both industry and academia have proposed many different neuromorphic architectures to execute applications that are designed with Spiking Neural Network (SNN). Consequently, there is a growing need for an extensible simulation framework that can perform architectural explorations with SNNs, including both platform-based design of today's hardware, and hardware-software co-design and design-technology co-optimization of the future. We present NeuroXplorer, a fast and extensible framework that is based on a generalized template for modeling a neuromorphic architecture that can be infused with the specific details of a given hardware and/or technology. NeuroXplorer can perform both low-level cycle-accurate architectural simulations and high-level analysis with data-flow abstractions. NeuroXplorer's optimization engine can incorporate hardware-oriented metrics such as energy, throughput, and latency, as well as SNN-oriented metrics such as inter-spike interval distortion and spike disorder, which directly impact SNN performance. We demonstrate the architectural exploration capabilities of NeuroXplorer through case studies with many state-of-the-art machine learning models.

Twisha Titirsha, Shihao Song, Anup Das et al.

Neuromorphic computing systems are embracing memristors to implement high density and low power synaptic storage as crossbar arrays in hardware. These systems are energy efficient in executing Spiking Neural Networks (SNNs). We observe that long bitlines and wordlines in a memristive crossbar are a major source of parasitic voltage drops, which create current asymmetry. Through circuit simulations, we show the significant endurance variation that results from this asymmetry. Therefore, if the critical memristors (ones with lower endurance) are overutilized, they may lead to a reduction of the crossbar's lifetime. We propose eSpine, a novel technique to improve lifetime by incorporating the endurance variation within each crossbar in mapping machine learning workloads, ensuring that synapses with higher activation are always implemented on memristors with higher endurance, and vice versa. eSpine works in two steps. First, it uses the Kernighan-Lin Graph Partitioning algorithm to partition a workload into clusters of neurons and synapses, where each cluster can fit in a crossbar. Second, it uses an instance of Particle Swarm Optimization (PSO) to map clusters to tiles, where the placement of synapses of a cluster to memristors of a crossbar is performed by analyzing their activation within the workload. We evaluate eSpine for a state-of-the-art neuromorphic hardware model with phase-change memory (PCM)-based memristors. Using 10 SNN workloads, we demonstrate a significant improvement in the effective lifetime.

Amir M. Rahmani, Jocelyn Lai, Salar Jafarlou et al.

Traditionally, the regime of mental healthcare has followed an episodic psychotherapy model wherein patients seek care from a provider through a prescribed treatment plan developed over multiple provider visits. Recent advances in wearable and mobile technology have generated increased interest in digital mental healthcare that enables individuals to address episodic mental health symptoms. However, these efforts are typically reactive and symptom-focused and do not provide comprehensive, wrap-around, customized treatments that capture an individual's holistic mental health model as it unfolds over time. Recognizing that each individual is unique, we present the notion of Personalized Mental Health Navigation (MHN): a therapist-in-the-loop, cybernetic goal-based system that deploys a continuous cyclic loop of measurement, estimation, guidance, to steer the individual's mental health state towards a healthy zone. We outline the major components of MHN that is premised on the development of an individual's personal mental health state, holistically represented by a high-dimensional cover of multiple knowledge layers such as emotion, biological patterns, sociology, behavior, and cognition. We demonstrate the feasibility of the personalized MHN approach via a 12-month pilot case study for holistic stress management in college students and highlight an instance of a therapist-in-the-loop intervention using MHN for monitoring, estimating, and proactively addressing moderately severe depression over a sustained period of time. We believe MHN paves the way to transform mental healthcare from the current passive, episodic, reactive process (where individuals seek help to address symptoms that have already manifested) to a continuous and navigational paradigm that leverages a personalized model of the individual, promising to deliver timely interventions to individuals in a holistic manner.

Milad Asgari Mehrabadi, Seyed Amir Hossein Aqajari, Iman Azimi et al.

The world has been affected by COVID-19 coronavirus. At the time of this study, the number of infected people in the United States is the highest globally (7.9 million infections). Within the infected population, patients diagnosed with acute respiratory distress syndrome (ARDS) are in more life-threatening circumstances, resulting in severe respiratory system failure. Various studies have investigated the infections to COVID-19 and ARDS by monitoring laboratory metrics and symptoms. Unfortunately, these methods are merely limited to clinical settings, and symptom-based methods are shown to be ineffective. In contrast, vital signs (e.g., heart rate) have been utilized to early-detect different respiratory diseases in ubiquitous health monitoring. We posit that such biomarkers are informative in identifying ARDS patients infected with COVID-19. In this study, we investigate the behavior of COVID-19 on ARDS patients by utilizing simple vital signs. We analyze the long-term daily logs of blood pressure and heart rate associated with 70 ARDS patients admitted to five University of California academic health centers (containing 42506 samples for each vital sign) to distinguish subjects with COVID-19 positive and negative test results. In addition to the statistical analysis, we develop a deep neural network model to extract features from the longitudinal data. Using only the first eight days of the data, our deep learning model is able to achieve 78.79% accuracy to classify the vital signs of ARDS patients infected with COVID-19 versus other ARDS diagnosed patients.

Wonseok Jang, Hansaem Jeong, Kyungtae Kang et al.

For realizing safe autonomous driving, the end-to-end delays of real-time object detection systems should be thoroughly analyzed and minimized. However, despite recent development of neural networks with minimized inference delays, surprisingly little attention has been paid to their end-to-end delays from an object's appearance until its detection is reported. With this motivation, this paper aims to provide more comprehensive understanding of the end-to-end delay, through which precise best- and worst-case delay predictions are formulated, and three optimization methods are implemented: (i) on-demand capture, (ii) zero-slack pipeline, and (iii) contention-free pipeline. Our experimental results show a 76% reduction in the end-to-end delay of Darknet YOLO (You Only Look Once) v3 (from 1070 ms to 261 ms), thereby demonstrating the great potential of exploiting the end-to-end delay analysis for autonomous driving. Furthermore, as we only modify the system architecture and do not change the neural network architecture itself, our approach incurs no penalty on the detection accuracy.

Adarsha Balaji, Shihao Song, Anup Das et al.

With growing model complexity, mapping Spiking Neural Network (SNN)-based applications to tile-based neuromorphic hardware is becoming increasingly challenging. This is because the synaptic storage resources on a tile, viz. a crossbar, can accommodate only a fixed number of pre-synaptic connections per post-synaptic neuron. For complex SNN models that have many pre-synaptic connections per neuron, some connections may need to be pruned after training to fit onto the tile resources, leading to a loss in model quality, e.g., accuracy. In this work, we propose a novel unrolling technique that decomposes a neuron function with many pre-synaptic connections into a sequence of homogeneous neural units, where each neural unit is a function computation node, with two pre-synaptic connections. This spatial decomposition technique significantly improves crossbar utilization and retains all pre-synaptic connections, resulting in no loss of the model quality derived from connection pruning. We integrate the proposed technique within an existing SNN mapping framework and evaluate it using machine learning applications on the DYNAP-SE state-of-the-art neuromorphic hardware. Our results demonstrate an average 60% lower crossbar requirement, 9x higher synapse utilization, 62% lower wasted energy on the hardware, and between 0.8% and 4.6% increase in model quality.

Rosario Cammarota, Matthias Schunter, Anand Rajan et al.

In this work, we provide an industry research view for approaching the design, deployment, and operation of trustworthy Artificial Intelligence (AI) inference systems. Such systems provide customers with timely, informed, and customized inferences to aid their decision, while at the same time utilizing appropriate security protection mechanisms for AI models. Additionally, such systems should also use Privacy-Enhancing Technologies (PETs) to protect customers' data at any time. To approach the subject, we start by introducing current trends in AI inference systems. We continue by elaborating on the relationship between Intellectual Property (IP) and private data protection in such systems. Regarding the protection mechanisms, we survey the security and privacy building blocks instrumental in designing, building, deploying, and operating private AI inference systems. For example, we highlight opportunities and challenges in AI systems using trusted execution environments combined with more recent advances in cryptographic techniques to protect data in use. Finally, we outline areas of further development that require the global collective attention of industry, academia, and government researchers to sustain the operation of trustworthy AI inference systems.

Milad Asgari Mehrabadi, Nikil Dutt, Amir M. Rahmani

The COVID-19 coronavirus pandemic has affected virtually every region of the globe. At the time of conducting this study, the number of daily cases in the United States is more than any other country, and the trend is increasing in most of its states. Google trends provide public interest in various topics during different periods. Analyzing these trends using data mining methods might provide useful insights and observations regarding the COVID-19 outbreak. The objective of this study was to consider the predictive ability of different search terms (i.e., bars and restaurants) with regards to the increase of daily cases in the US. We considered the causation of two different search query trends, namely restaurant and bars, on daily positive cases in top-10 states/territories of the United States with the highest and lowest daily new positive cases. In addition, to measure the linear relation of different trends, we used Pearson correlation. Our results showed for states/territories with higher numbers of daily cases, the historical trends in search queries related to bars and restaurants, which mainly happened after re-opening, significantly affect the daily new cases, on average. California, for example, had most searches for restaurants on June 7th, 2020, which affected the number of new cases within two weeks after the peak with the P-value of .004 for Granger's causality test. Although a limited number of search queries were considered, Google search trends for restaurants and bars showed a significant effect on daily new cases for regions with higher numbers of daily new cases in the United States. We showed that such influential search trends could be used as additional information for prediction tasks in new cases of each region. This prediction can help healthcare leaders manage and control the impact of COVID-19 outbreaks on society and be prepared for the outcomes.

Adarsha Balaji, Shihao Song, Anup Das et al.

Neuromorphic hardware with non-volatile memory (NVM) can implement machine learning workload in an energy-efficient manner. Unfortunately, certain NVMs such as phase change memory (PCM) require high voltages for correct operation. These voltages are supplied from an on-chip charge pump. If the charge pump is activated too frequently, its internal CMOS devices do not recover from stress, accelerating their aging and leading to negative bias temperature instability (NBTI) generated defects. Forcefully discharging the stressed charge pump can lower the aging rate of its CMOS devices, but makes the neuromorphic hardware unavailable to perform computations while its charge pump is being discharged. This negatively impacts performance such as latency and accuracy of the machine learning workload being executed. In this paper, we propose a novel framework to exploit workload-specific performance and lifetime trade-offs in neuromorphic computing. Our framework first extracts the precise times at which a charge pump in the hardware is activated to support neural computations within a workload. This timing information is then used with a characterized NBTI reliability model to estimate the charge pump's aging during the workload execution. We use our framework to evaluate workload-specific performance and reliability impacts of using 1) different SNN mapping strategies and 2) different charge pump discharge strategies. We show that our framework can be used by system designers to explore performance and reliability trade-offs early in the design of neuromorphic hardware such that appropriate reliability-oriented design margins can be set.

Adarsha Balaji, Anup Das, Yuefeng Wu et al.

Neuromorphic hardware platforms implement biological neurons and synapses to execute spiking neural networks (SNNs) in an energy-efficient manner. We present SpiNeMap, a design methodology to map SNNs to crossbar-based neuromorphic hardware, minimizing spike latency and energy consumption. SpiNeMap operates in two steps: SpiNeCluster and SpiNePlacer. SpiNeCluster is a heuristic-based clustering technique to partition SNNs into clusters of synapses, where intracluster local synapses are mapped within crossbars of the hardware and inter-cluster global synapses are mapped to the shared interconnect. SpiNeCluster minimizes the number of spikes on global synapses, which reduces spike congestion on the shared interconnect, improving application performance. SpiNePlacer then finds the best placement of local and global synapses on the hardware using a meta-heuristic-based approach to minimize energy consumption and spike latency. We evaluate SpiNeMap using synthetic and realistic SNNs on the DynapSE neuromorphic hardware. We show that SpiNeMap reduces average energy consumption by 45% and average spike latency by 21%, compared to state-of-the-art techniques.

Delaram Amiri, Arman Anzanpour, Iman Azimi et al.

Recent advances in the Internet of Things (IoT) technologies have enabled the use of wearables for remote patient monitoring. Wearable sensors capture the patient's vital signs, and provide alerts or diagnosis based on the collected data. Unfortunately, wearables typically have limited energy and computational capacity, making their use challenging for healthcare applications where monitoring must continue uninterrupted long time, without the need to charge or change the battery. Fog computing can alleviate this problem by offloading computationally intensive tasks from the sensor layer to higher layers, thereby not only meeting the sensors' limited computational capacity but also enabling the use of local closed-loop energy optimization algorithms to increase the battery life.

Georgios Detorakis, Sadique Sheik, Charles Augustine et al.

Embedded, continual learning for autonomous and adaptive behavior is a key application of neuromorphic hardware. However, neuromorphic implementations of embedded learning at large scales that are both flexible and efficient have been hindered by a lack of a suitable algorithmic framework. As a result, the most neuromorphic hardware is trained off-line on large clusters of dedicated processors or GPUs and transferred post hoc to the device. We address this by introducing the neural and synaptic array transceiver (NSAT), a neuromorphic computational framework facilitating flexible and efficient embedded learning by matching algorithmic requirements and neural and synaptic dynamics. NSAT supports event-driven supervised, unsupervised and reinforcement learning algorithms including deep learning. We demonstrate the NSAT in a wide range of tasks, including the simulation of Mihalas-Niebur neuron, dynamic neural fields, event-driven random back-propagation for event-based deep learning, event-based contrastive divergence for unsupervised learning, and voltage-based learning rules for sequence learning. We anticipate that this contribution will establish the foundation for a new generation of devices enabling adaptive mobile systems, wearable devices, and robots with data-driven autonomy.

Anup Das, Paruthi Pradhapan, Willemijn Groenendaal et al.

Heart-rate estimation is a fundamental feature of modern wearable devices. In this paper we propose a machine intelligent approach for heart-rate estimation from electrocardiogram (ECG) data collected using wearable devices. The novelty of our approach lies in (1) encoding spatio-temporal properties of ECG signals directly into spike train and using this to excite recurrently connected spiking neurons in a Liquid State Machine computation model; (2) a novel learning algorithm; and (3) an intelligently designed unsupervised readout based on Fuzzy c-Means clustering of spike responses from a subset of neurons (Liquid states), selected using particle swarm optimization. Our approach differs from existing works by learning directly from ECG signals (allowing personalization), without requiring costly data annotations. Additionally, our approach can be easily implemented on state-of-the-art spiking-based neuromorphic systems, offering high accuracy, yet significantly low energy footprint, leading to an extended battery life of wearable devices. We validated our approach with CARLsim, a GPU accelerated spiking neural network simulator modeling Izhikevich spiking neurons with Spike Timing Dependent Plasticity (STDP) and homeostatic scaling. A range of subjects are considered from in-house clinical trials and public ECG databases. Results show high accuracy and low energy footprint in heart-rate estimation across subjects with and without cardiac irregularities, signifying the strong potential of this approach to be integrated in future wearable devices.