Search

While large language models (LLMs) dominate the AI landscape, Small-scale large Language Models (SLMs) are gaining attention due to cost and efficiency demands from consumers. However, there is limited research on the training behavior and computational requirements of SLMs. In this study, we explore the computational bottlenecks of training SLMs (up to 2B parameters) by examining the effects of various hyperparameters and configurations, including GPU type, batch size, model size, communication protocol, attention type, and the number of GPUs. We assess these factors on popular cloud services using metrics such as loss per dollar and tokens per second. Our findings aim to support the broader adoption and optimization of language model training for low-resource AI research institutes., Comment: 8 pages, 4 figures

Eyelid shape is integral to identity and likeness in human facial modeling. Human eyelids are diverse in appearance with varied skin fold and epicanthal fold morphology between individuals. Existing parametric face models express eyelid shape variation to an extent, but do not preserve sufficient likeness across a diverse range of individuals. We propose a new definition of eyelid fold consistency and implement geometric processing techniques to model diverse eyelid shapes in a unified topology. Using this method we reprocess data used to train a parametric face model and demonstrate significant improvements in face-related machine learning tasks.

Healthcare industries face challenges when experiencing rare diseases due to limited samples. Artificial Intelligence (AI) communities overcome this situation to create synthetic data which is an ethical and privacy issue in the medical domain. This research introduces the CAT-U-Net framework as a new approach to overcome these limitations, which enhances feature extraction from medical images without the need for large datasets. The proposed framework adds an extra concatenation layer with downsampling parts, thereby improving its ability to learn from limited data while maintaining patient privacy. To validate, the proposed framework's robustness, different medical conditioning datasets were utilized including COVID-19, brain tumors, and wrist fractures. The framework achieved nearly 98% reconstruction accuracy, with a Dice coefficient close to 0.946. The proposed CAT-U-Net has the potential to make a big difference in medical image diagnostics in settings with limited data., Comment: 7 pages, 7 figures

By 2025, the internet of things (IoT) is projected to connect over 75 billion devices globally, fundamentally altering how we interact with our environments in both urban and rural settings. However, IoT device security remains challenging, particularly in the authentication process. Traditional cryptographic methods often struggle with the constraints of IoT devices, such as limited computational power and storage. This paper considers physical unclonable functions (PUFs) as robust security solutions, utilizing their inherent physical uniqueness to authenticate devices securely. However, traditional PUF systems are vulnerable to machine learning (ML) attacks and burdened by large datasets. Our proposed solution introduces a lightweight PUF mechanism, called LPUF-AuthNet, combining tandem neural networks (TNN) with a split learning (SL) paradigm. The proposed approach provides scalability, supports mutual authentication, and enhances security by resisting various types of attacks, paving the way for secure integration into future 6G technologies., Comment: Accepted to Proc. IEEE Globecom 2024

We tackle the problem of highly-accurate, holistic performance capture for the face, body and hands simultaneously. Motion-capture technologies used in film and game production typically focus only on face, body or hand capture independently, involve complex and expensive hardware and a high degree of manual intervention from skilled operators. While machine-learning-based approaches exist to overcome these problems, they usually only support a single camera, often operate on a single part of the body, do not produce precise world-space results, and rarely generalize outside specific contexts. In this work, we introduce the first technique for marker-free, high-quality reconstruction of the complete human body, including eyes and tongue, without requiring any calibration, manual intervention or custom hardware. Our approach produces stable world-space results from arbitrary camera rigs as well as supporting varied capture environments and clothing. We achieve this through a hybrid approach that leverages machine learning models trained exclusively on synthetic data and powerful parametric models of human shape and motion. We evaluate our method on a number of body, face and hand reconstruction benchmarks and demonstrate state-of-the-art results that generalize on diverse datasets.

6G networks aim to enable applications like autonomous driving by providing complementary localization services through key technologies such as non-terrestrial networks (NTNs) with low Earth orbit (LEO) satellites and reconfigurable intelligent surfaces (RIS). Prior research in 6G localization using single LEO, multi-LEO, and multi-LEO multi-RIS setups has limitations: single LEO lacks the required accuracy, while multi-LEO/RIS setups demand many visible satellites and RISs, which is not always feasible in practice. This paper explores the novel problem of localization with a single LEO satellite and a single RIS, bridging these research areas. We present a comprehensive signal model accounting for user carrier frequency offset (CFO), clock bias, and fast and slow Doppler effects. Additionally, we derive a low-complexity estimator that achieves theoretical bounds at high signal-to-noise ratios (SNR). Our results demonstrate the feasibility and accuracy of RIS-aided single-LEO localization in 6G networks and highlight potential research directions., Comment: Accepted for publication in IEEE GLOBECOM Workshops 2024

Modern Systems-on-Chip (SoCs) incorporate built-in self-test (BIST) modules deeply integrated into the device's intellectual property (IP) blocks. Such modules handle hardware faults and defects during device operation. As such, BIST results potentially reveal the internal structure and state of the device under test (DUT) and hence open attack vectors. So-called result compaction can overcome this vulnerability by hiding the BIST chain structure but introduces the issues of aliasing and invalid signatures. Software-BIST provides a flexible solution, that can tackle these issues, but suffers from limited observability and fault coverage. In this paper, we hence introduce a low-overhead software/hardware hybrid approach that overcomes the mentioned limitations. It relies on (a) keyed-hash message authentication code (KMAC) available on the SoC providing device-specific secure and valid signatures with zero aliasing and (b) the SoC processor for test scheduling hence increasing DUT availability. The proposed approach offers both on-chip- and remote-testing capabilities. We showcase a RISC-V-based SoC to demonstrate our approach, discussing system overhead and resulting compaction rates.

While large language models (LLMs) have exhibited impressive conversational capabilities, their proficiency in delivering personalized responses remains unclear. Although recent benchmarks automatically evaluate persona consistency in role-playing contexts using LLM-based judgment, the evaluation of personalization in response generation remains underexplored. To address this gap, we present a new benchmark, PersoBench, to evaluate the personalization ability of LLMs in persona-aware dialogue generation within a zero-shot setting. We assess the performance of three open-source and three closed-source LLMs using well-known datasets and a range of metrics. Our analysis, conducted on three well-known persona-aware datasets, evaluates multiple dimensions of response quality, including fluency, diversity, coherence, and personalization, across both standard and chain-of-thought prompting methods. Our findings reveal that while LLMs excel at generating fluent and diverse responses, they are far from satisfactory in delivering personalized and coherent responses considering both the conversation context and the provided personas. Our benchmark implementation is available at https://github.com/salehafzoon/PersoBench.

Monocular Depth and Surface Normals Estimation (MDSNE) is crucial for tasks such as 3D reconstruction, autonomous navigation, and underwater exploration. Current methods rely either on discriminative models, which struggle with transparent or reflective surfaces, or generative models, which, while accurate, are computationally expensive. This paper presents a novel deep learning model for MDSNE, specifically tailored for underwater environments, using a hybrid architecture that integrates Convolutional Neural Networks (CNNs) with Transformers, leveraging the strengths of both approaches. Training effective MDSNE models is often hampered by noisy real-world datasets and the limited generalization of synthetic datasets. To address this, we generate pseudo-labeled real data using multiple pre-trained MDSNE models. To ensure the quality of this data, we propose the Depth Normal Evaluation and Selection Algorithm (DNESA), which evaluates and selects the most reliable pseudo-labeled samples using domain-specific metrics. A lightweight student model is then trained on this curated dataset. Our model reduces parameters by 90% and training costs by 80%, allowing real-time 3D perception on resource-constrained devices. Key contributions include: a novel and efficient MDSNE model, the DNESA algorithm, a domain-specific data pipeline, and a focus on real-time performance and scalability. Designed for real-world underwater applications, our model facilitates low-cost deployments in underwater robots and autonomous vehicles, bridging the gap between research and practical implementation., Comment: 18 pages, 6 figures, 8 tables. Submitted to Elsevier

Fire hazards are extremely dangerous, particularly in sectors such as the transportation industry, where political unrest increases the likelihood of their occurrence. By employing IP cameras to facilitate the setup of fire detection systems on transport vehicles, losses from fire events may be prevented proactively. However, the development of lightweight fire detection models is required due to the computational constraints of the embedded systems within these cameras. We introduce FireLite, a low-parameter convolutional neural network (CNN) designed for quick fire detection in contexts with limited resources, in response to this difficulty. With an accuracy of 98.77\%, our model -- which has just 34,978 trainable parameters achieves remarkable performance numbers. It also shows a validation loss of 8.74 and peaks at 98.77 for precision, recall, and F1-score measures. Because of its precision and efficiency, FireLite is a promising solution for fire detection in resource-constrained environments.

Cancer patients often struggle to transition swiftly to treatment due to limited institutional resources, lack of sophisticated professional guidance, and low health literacy. The emergence of Large Language Models (LLMs) offers new opportunities for such patients to access the wealth of existing patient education materials. The current paper presents the development process for an LLM-based chatbot focused on prostate cancer education, including needs assessment, co-design, and usability studies. The resulting application, MedEduChat, integrates with patients' electronic health record data and features a closed-domain, semi-structured, patient-centered approach to address real-world needs. This paper contributes to the growing field of patient-LLM interaction by demonstrating the potential of LLM-based chatbots to enhance prostate cancer patient education and by offering co-design guidelines for future LLM-based healthcare downstream applications.

Reward models (RMs) play a pivotal role in aligning large language models (LLMs) with human preferences. However, traditional RM training, which relies on response pairs tied to specific prompts, struggles to disentangle prompt-driven preferences from prompt-independent artifacts, such as response length and format. In this work, we expose a fundamental limitation of current RM training methods, where RMs fail to effectively distinguish between contextual signals and irrelevant artifacts when determining preferences. To address this, we introduce a causal framework that learns preferences independent of these artifacts and propose a novel data augmentation technique designed to eliminate them. Extensive experiments show that our approach successfully filters out undesirable artifacts, yielding a more robust reward model (RRM). Our RRM improves the performance of a pairwise reward model trained on Gemma-2-9b-it, on RewardBench, increasing accuracy from 80.61% to 84.15%. Additionally, we train two DPO policies using both the RM and RRM, demonstrating that the RRM significantly enhances DPO-aligned policies, improving MT-Bench scores from 7.27 to 8.31 and length-controlled win-rates in AlpacaEval-2 from 33.46% to 52.49%., Comment: Accepted in ICLR 2025

Weakly supervised video anomaly detection (WS-VAD) is a crucial area in computer vision for developing intelligent surveillance systems. This system uses three feature streams: RGB video, optical flow, and audio signals, where each stream extracts complementary spatial and temporal features using an enhanced attention module to improve detection accuracy and robustness. In the first stream, we employed an attention-based, multi-stage feature enhancement approach to improve spatial and temporal features from the RGB video where the first stage consists of a ViT-based CLIP module, with top-k features concatenated in parallel with I3D and Temporal Contextual Aggregation (TCA) based rich spatiotemporal features. The second stage effectively captures temporal dependencies using the Uncertainty-Regulated Dual Memory Units (UR-DMU) model, which learns representations of normal and abnormal data simultaneously, and the third stage is employed to select the most relevant spatiotemporal features. The second stream extracted enhanced attention-based spatiotemporal features from the flow data modality-based feature by taking advantage of the integration of the deep learning and attention module. The audio stream captures auditory cues using an attention module integrated with the VGGish model, aiming to detect anomalies based on sound patterns. These streams enrich the model by incorporating motion and audio signals often indicative of abnormal events undetectable through visual analysis alone. The concatenation of the multimodal fusion leverages the strengths of each modality, resulting in a comprehensive feature set that significantly improves anomaly detection accuracy and robustness across three datasets. The extensive experiment and high performance with the three benchmark datasets proved the effectiveness of the proposed system over the existing state-of-the-art system.

Human Activity Recognition (HAR) systems aim to understand human behaviour and assign a label to each action, attracting significant attention in computer vision due to their wide range of applications. HAR can leverage various data modalities, such as RGB images and video, skeleton, depth, infrared, point cloud, event stream, audio, acceleration, and radar signals. Each modality provides unique and complementary information suited to different application scenarios. Consequently, numerous studies have investigated diverse approaches for HAR using these modalities. This paper presents a comprehensive survey of the latest advancements in HAR from 2014 to 2024, focusing on machine learning (ML) and deep learning (DL) approaches categorized by input data modalities. We review both single-modality and multi-modality techniques, highlighting fusion-based and co-learning frameworks. Additionally, we cover advancements in hand-crafted action features, methods for recognizing human-object interactions, and activity detection. Our survey includes a detailed dataset description for each modality and a summary of the latest HAR systems, offering comparative results on benchmark datasets. Finally, we provide insightful observations and propose effective future research directions in HAR.

Considering the growing prominence of production-level AI and the threat of adversarial attacks that can evade a model at run-time, evaluating the robustness of models to these evasion attacks is of critical importance. Additionally, testing model changes likely means deploying the models to (e.g. a car or a medical imaging device), or a drone to see how it affects performance, making un-tested changes a public problem that reduces development speed, increases cost of development, and makes it difficult (if not impossible) to parse cause from effect. In this work, we used survival analysis as a cloud-native, time-efficient and precise method for predicting model performance in the presence of adversarial noise. For neural networks in particular, the relationships between the learning rate, batch size, training time, convergence time, and deployment cost are highly complex, so researchers generally rely on benchmark datasets to assess the ability of a model to generalize beyond the training data. To address this, we propose using accelerated failure time models to measure the effect of hardware choice, batch size, number of epochs, and test-set accuracy by using adversarial attacks to induce failures on a reference model architecture before deploying the model to the real world. We evaluate several GPU types and use the Tree Parzen Estimator to maximize model robustness and minimize model run-time simultaneously. This provides a way to evaluate the model and optimise it in a single step, while simultaneously allowing us to model the effect of model parameters on training time, prediction time, and accuracy. Using this technique, we demonstrate that newer, more-powerful hardware does decrease the training time, but with a monetary and power cost that far outpaces the marginal gains in accuracy.

We establish empirical risk minimization principles for active learning by deriving a family of upper bounds on the generalization error. Aligning with empirical observations, the bounds suggest that superior query algorithms can be obtained by combining both informativeness and representativeness query strategies, where the latter is assessed using integral probability metrics. To facilitate the use of these bounds in application, we systematically link diverse active learning scenarios, characterized by their loss functions and hypothesis classes to their corresponding upper bounds. Our results show that regularization techniques used to constraint the complexity of various hypothesis classes are sufficient conditions to ensure the validity of the bounds. The present work enables principled construction and empirical quality-evaluation of query algorithms in active learning.

As network services progress and mobile and IoT environments expand, numerous security concerns have surfaced for spectrum access systems. The omnipresent risk of Denial-of-Service (DoS) attacks and raising concerns about user privacy (e.g., location privacy, anonymity) are among such cyber threats. These security and privacy risks increase due to the threat of quantum computers that can compromise long-term security by circumventing conventional cryptosystems and increasing the cost of countermeasures. While some defense mechanisms exist against these threats in isolation, there is a significant gap in the state of the art on a holistic solution against DoS attacks with privacy and anonymity for spectrum management systems, especially when post-quantum (PQ) security is in mind. In this paper, we propose a new cybersecurity framework PACDoSQ, which is (to the best of our knowledge) the first to offer location privacy and anonymity for spectrum management with counter DoS and PQ security simultaneously. Our solution introduces the private spectrum bastion (database) concept to exploit existing architectural features of spectrum management systems and then synergizes them with multi-server private information retrieval and PQ-secure Tor to guarantee a location-private and anonymous acquisition of spectrum information together with hash-based client-server puzzles for counter DoS. We prove that PACDoSQ achieves its security objectives, and show its feasibility via a comprehensive performance evaluation., Comment: Accepted at IEEE Military Communications Conference, 28 October 1 November 2024, Washington, DC, USA

Recent studies have shown that large language models' (LLMs) mathematical problem-solving capabilities can be enhanced by integrating external tools, such as code interpreters, and employing multi-turn Chain-of-Thought (CoT) reasoning. While current methods focus on synthetic data generation and Supervised Fine-Tuning (SFT), this paper studies the complementary direct preference learning approach to further improve model performance. However, existing direct preference learning algorithms are originally designed for the single-turn chat task, and do not fully address the complexities of multi-turn reasoning and external tool integration required for tool-integrated mathematical reasoning tasks. To fill in this gap, we introduce a multi-turn direct preference learning framework, tailored for this context, that leverages feedback from code interpreters and optimizes trajectory-level preferences. This framework includes multi-turn DPO and multi-turn KTO as specific implementations. The effectiveness of our framework is validated through training of various language models using an augmented prompt set from the GSM8K and MATH datasets. Our results demonstrate substantial improvements: a supervised fine-tuned Gemma-1.1-it-7B model's performance increased from 77.5% to 83.9% on GSM8K and from 46.1% to 51.2% on MATH. Similarly, a Gemma-2-it-9B model improved from 84.1% to 86.3% on GSM8K and from 51.0% to 54.5% on MATH., Comment: A multi-turn direct preference learning framework for tool-integrated reasoning tasks

This study evaluates the accessibility of public EV charging stations in the Washington metropolitan area using a comprehensive measure that accounts for both destination-based and en route charging opportunities. By incorporating the full spectrum of daily travel patterns into the accessibility evaluation, our methodology offers a more realistic measure of charging opportunities than destination-based methods that prioritize proximity to residential locations. Results from spatial autocorrelation analysis indicate that conventional accessibility assessments often overestimate the availability of infrastructure in central urban areas and underestimate it in peripheral commuting zones, potentially leading to misallocated resources. By highlighting significant clusters of high-access and low-access areas, our approach identifies spatial inequalities in infrastructure distribution and provides insights into areas requiring targeted interventions. This study underscores the importance of incorporating daily mobility patterns into urban planning to ensure equitable access to EV charging infrastructure and suggests a framework that other regions could adopt to enhance sustainable transportation networks and support equitable urban development.

Hand gesture-based sign language recognition (SLR) is one of the most advanced applications of machine learning, and computer vision uses hand gestures. Although, in the past few years, many researchers have widely explored and studied how to address BSL problems, specific unaddressed issues remain, such as skeleton and transformer-based BSL recognition. In addition, the lack of evaluation of the BSL model in various concealed environmental conditions can prove the generalized property of the existing model by facing daily life signs. As a consequence, existing BSL recognition systems provide a limited perspective of their generalisation ability as they are tested on datasets containing few BSL alphabets that have a wide disparity in gestures and are easy to differentiate. To overcome these limitations, we propose a spatial-temporal attention-based BSL recognition model considering hand joint skeletons extracted from the sequence of images. The main aim of utilising hand skeleton-based BSL data is to ensure the privacy and low-resolution sequence of images, which need minimum computational cost and low hardware configurations. Our model captures discriminative structural displacements and short-range dependency based on unified joint features projected onto high-dimensional feature space. Specifically, the use of Separable TCN combined with a powerful multi-head spatial-temporal attention architecture generated high-performance accuracy. The extensive experiments with a proposed dataset and two benchmark BSL datasets with a wide range of evaluations, such as intra- and inter-dataset evaluation settings, demonstrated that our proposed models achieve competitive performance with extremely low computational complexity and run faster than existing models.

Developmental dysgraphia is a neurological disorder that hinders children's writing skills. In recent years, researchers have increasingly explored machine learning methods to support the diagnosis of dysgraphia based on offline and online handwriting. In most previous studies, the two types of handwriting have been analysed separately, which does not necessarily lead to promising results. In this way, the relationship between online and offline data cannot be explored. To address this limitation, we propose a novel multimodal machine learning approach utilizing both online and offline handwriting data. We created a new dataset by transforming an existing online handwritten dataset, generating corresponding offline handwriting images. We considered only different types of word data (simple word, pseudoword & difficult word) in our multimodal analysis. We trained SVM and XGBoost classifiers separately on online and offline features as well as implemented multimodal feature fusion and soft-voted ensemble. Furthermore, we proposed a novel ensemble with conditional feature fusion method which intelligently combines predictions from online and offline classifiers, selectively incorporating feature fusion when confidence scores fall below a threshold. Our novel approach achieves an accuracy of 88.8%, outperforming SVMs for single modalities by 12-14%, existing methods by 8-9%, and traditional multimodal approaches (soft-vote ensemble and feature fusion) by 3% and 5%, respectively. Our methodology contributes to the development of accurate and efficient dysgraphia diagnosis tools, requiring only a single instance of multimodal word/pseudoword data to determine the handwriting impairment. This work highlights the potential of multimodal learning in enhancing dysgraphia diagnosis, paving the way for accessible and practical diagnostic tools.

Surface electromyography (EMG) serves as a pivotal tool in hand gesture recognition and human-computer interaction, offering a non-invasive means of signal acquisition. This study presents a novel methodology for classifying hand gestures using EMG signals. To address the challenges associated with feature extraction where, we explored 23 distinct morphological, time domain and frequency domain feature extraction techniques. However, the substantial size of the features may increase the computational complexity issues that can hinder machine learning algorithm performance. We employ an efficient feature selection approach, specifically an extra tree classifier, to mitigate this. The selected potential feature fed into the various machine learning-based classification algorithms where our model achieved 97.43\% accuracy with the KNN algorithm and selected feature. By leveraging a comprehensive feature extraction and selection strategy, our methodology enhances the accuracy and usability of EMG-based hand gesture recognition systems. The higher performance accuracy proves the effectiveness of the proposed model over the existing system. \keywords{EMG signal, machine learning approach, hand gesture recognition.

The prevention of falls is paramount in modern healthcare, particularly for the elderly, as falls can lead to severe injuries or even fatalities. Additionally, the growing incidence of falls among the elderly, coupled with the urgent need to prevent suicide attempts resulting from medication overdose, underscores the critical importance of accurate and efficient fall detection methods. In this scenario, a computer-aided fall detection system is inevitable to save elderly people's lives worldwide. Many researchers have been working to develop fall detection systems. However, the existing fall detection systems often struggle with issues such as unsatisfactory performance accuracy, limited robustness, high computational complexity, and sensitivity to environmental factors due to a lack of effective features. In response to these challenges, this paper proposes a novel three-stream spatial-temporal feature-based fall detection system. Our system incorporates joint skeleton-based spatial and temporal Graph Convolutional Network (GCN) features, joint motion-based spatial and temporal GCN features, and residual connections-based features. Each stream employs adaptive graph-based feature aggregation and consecutive separable convolutional neural networks (Sep-TCN), significantly reducing computational complexity and model parameters compared to prior systems. Experimental results across multiple datasets demonstrate the superior effectiveness and efficiency of our proposed system, with accuracies of 99.51\%, 99.15\%, 99.79\% and 99.85 \% achieved on the ImViA, UR-Fall, Fall-UP and FU-Kinect datasets, respectively. The remarkable performance of our system highlights its superiority, efficiency, and generalizability in real-world fall detection scenarios, offering significant advancements in healthcare and societal well-being.

We show that randomness in quadratic bosonic Hamiltonians results in certain information scrambling diagnostics, mirroring those in chaotic systems. Specifically, for initial Gaussian states, we observe the disappearance of the memory effect in the entanglement dynamics of disjoint blocks and the negative values of tripartite mutual information. We also find that the spectral form factor for these integrable systems exhibits a ramp. However, in contrast to chaotic systems, the ramp is nonlinear, and the out-of-time-ordered correlators display power-law growth for certain operators and Gaussian dynamics. These findings indicate that information scrambling driven by randomness is distinct from quantum chaos. Moreover, our results provide insight into the dynamics of Gaussian states in continuous-variable systems, which are useful and available resources for quantum information processing., Comment: 8+5 pages, 4+2 figures, v2: references updated + presentation improved + added discussion about adding vacuum noise