Prior Works Research Articles

VSPIM: SRAM Processing-in-Memory DNN Acceleration via Vector-Scalar Operations

Processing-in-Memory (PIM) has been widely explored for accelerating data-intensive machine learning computation that mainly consists of general-matrix-multiplication (GEMM), by mitigating the burden of data movements and exploiting the ultra-high memory parallelism. The two mainstreams of PIM, the analog- and digital-type, have both been exploited in accelerating machine learning workloads by numerous outstanding prior works. Currently, the digital-PIM is increasingly favored due to the broader computing support and the avoidance of errors caused by intrinsic non-idealities, e.g., process variation. Nevertheless, it still lacks further optimization considering the characteristics of the GEMM computation, including better efficient data layout and scheduling, and the ability to handle the sparsity of activations at the bit-level. To boost the performance and efficiency of digital SRAM PIM, we propose the architecture called VSPIM that performs the computation in a bit-serial fashion, with unique support of vector-scalar computing pattern. The novelties of the VSPIM can be concluded as follows: 1) support bit-serial based scalar-vector computing via ingenious parallel bit-broadcasting; 2) refine the GEMM mapping strategy and computing pattern to enhance performance and efficiency; 3) powered by the introduced scalar-vector operation, the bit-sparsity of activation is leveraged to halt unnecessary computation to maximize efficiency and throughput. Our comprehensive evaluation shows that, compared to the state-of-the-art SRAM-based digital-PIM design (Neural Cache), VSPIM can significantly boost the performance and energy efficiency by up to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$8.87\times$</tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$4.81\times$</tex-math></inline-formula> respectively, with negligible area overhead, upon multiple representative neural networks.

A Fast FPGA Hardware Accelerator for Remote Heart Rate Detection Based on RGB Vision.

A fast hardware accelerator is created by this work via field programmable gate array (FPGA) to estimate heart rate (HR) through the video recorded by a RGB camera based on the technology of remote photoplethysmography (rPPG). The method of rPPG acquires physiological signals of a human body by analyzing the subtle color changes on the surface of the human skin. The hardware implementation of rPPG to estimate HR is proposed herein to aim for a much faster calculation speed than software for a number of applications, like heart failure pre-warning of an in-action athlete and drowsiness detection of a driver. In this accelerator, ICA (Independent Component Analysis) is used to recover the blood volume pulse from the raw signals of remote PPG, and then obtain the heart rate value. The architecture of the hardware circuit is described in Verilog HDL and verified by Quartus II, and also implemented in an Altera DE10-Standard FPGA board, which consists of image capture, heart rate algorithm and image display. A TRDB-D5M camera is utilized for image capture. Two experiments were conducted with image collecting duration of 16 seconds and 8 seconds respectively, and the commercial device Omron HEM-6111 was used as the golden value. The proposed system achieves an accuracy in (ME ± 1.96SD) of -0.76 ± 5.09 and -0.70 ± 8.71 bpm in the short periods of 16-second and 8-second versions, respectively, which outperforms all the reported prior works in combined computation time and accuracy.

In-Domain GAN Inversion for Faithful Reconstruction and Editability.

Generative Adversarial Networks (GANs) have significantly advanced image synthesis through mapping randomly sampled latent codes to high-fidelity synthesized images. However, applying well-trained GANs to real image editing remains challenging. A common solution is to find an approximate latent code that can adequately recover the input image to edit, which is also known as GAN inversion. To invert a GAN model, prior works typically focus on reconstructing the target image at the pixel level, yet few studies are conducted on whether the inverted result can well support manipulation at the semantic level. This work fills in this gap by proposing in-domain GAN inversion, which consists of a domain-guided encoder and a domain-regularized optimizer, to regularize the inverted code in the native latent space of the pre-trained GAN model. In this way, we manage to sufficiently reuse the knowledge learned by GANs for image reconstruction, facilitating a wide range of editing applications without any retraining. We further make comprehensive analyses on the effects of the encoder structure, the starting inversion point, as well as the inversion parameter space, and observe the trade-off between the reconstruction quality and the editing property. Such a trade-off sheds light on how a GAN model represents an image with various semantics encoded in the learned latent distribution.

Impression of excessive digital media utilization on psychological well-being: A case of Pakistan’s metropolitan cities

Digital media utilization assumes an imperative part in changing millennials’ way of life. The rapid growth in the usage of digital media may have an impact on the mental prosperity of users. The primary focus of this research is to examine how excessive usage of digital media affects its consumers' psychological well-being in Pakistan's major cities. In January this year, there were 72 million digital media users in Pakistan, equal to 32% of the total population, of which 28% of Pakistan is between the ages of 26-41 (Kemp, 2022). This is the most probable age group to be used and influenced by digital media. Based on prior works, it delivers a more inclusive study of the spectacle by the view of independent variables (social capital, social isolation, and smartphone addiction), dependent variable (psychological well-being), and one mediator (excessive digital media utilization). This study sheds light on the best ways to mitigate the negative effects of using digital media, which claims to have a clear adverse effect on mental health through social capital creation and management. Overall, this research provides important insights for academics, scholars, and society at large. This paper consists of a quantitative approach of 236 users from metropolitan cities in Pakistan with the help of structural equation modeling to test the projected propositions. The research study could be additionally upgraded if we center around more factors and other various areas of Pakistan for exploring the effect of excessive digital media utilization on mental prosperity.

Minimax Optimal Bandits for Heavy Tail Rewards.

Stochastic multiarmed bandits (stochastic MABs) are a problem of sequential decision-making with noisy rewards, where an agent sequentially chooses actions under unknown reward distributions to minimize cumulative regret. The majority of prior works on stochastic MABs assume that the reward distribution of each action has bounded supports or follows light-tailed distribution, i.e., sub-Gaussian distribution. However, in a variety of decision-making problems, the reward distributions follow a heavy-tailed distribution. In this regard, we consider stochastic MABs with heavy-tailed rewards, whose p th moment is bounded by a constant νp for . First, we provide theoretical analysis on sub-optimality of the existing exploration methods for heavy-tailed rewards where it has been proven that existing exploration methods do not guarantee a minimax optimal regret bound. Second, to achieve the minimax optimality under heavy-tailed rewards, we propose a minimax optimal robust upper confidence bound (MR-UCB) by providing tight confidence bound of a p -robust estimator. Furthermore, we also propose a minimax optimal robust adaptively perturbed exploration (MR-APE) which is a randomized version of MR-UCB. In particular, unlike the existing robust exploration methods, both proposed methods have no dependence on νp . Third, we provide the gap-dependent and independent regret bounds of proposed methods and prove that both methods guarantee the minimax optimal regret bound for a heavy-tailed stochastic MAB problem. The proposed methods are the first algorithm that theoretically guarantees the minimax optimality under heavy-tailed reward settings to the best of our knowledge. Finally, we demonstrate the superiority of the proposed methods in simulation with Pareto and Fréchet noises with respect to regrets.

Collaborative Tooth Motion Diffusion Model in Digital Orthodontics

Tooth motion generation is an essential task in digital orthodontic treatment for precise and quick dental healthcare, which aims to generate the whole intermediate tooth motion process given the initial pathological and target ideal tooth alignments. Most prior works for multi-agent motion planning problems usually result in complex solutions. Moreover, the occlusal relationship between upper and lower teeth is often overlooked. In this paper, we propose a collaborative tooth motion diffusion model. The critical insight is to remodel the problem as a diffusion process. In this sense, we model the whole tooth motion distribution with a diffusion model and transform the planning problem into a sampling process from this distribution. We design a tooth latent representation to provide accurate conditional guides consisting of two key components: the tooth frame represents the position and posture, and the tooth latent shape code represents the geometric morphology. Subsequently, we present a collaborative diffusion model to learn the multi-tooth motion distribution based on inter-tooth and occlusal constraints, which are implemented by graph structure and new loss functions, respectively. Extensive qualitative and quantitative experiments demonstrate the superiority of our framework in the application of orthodontics compared with state-of-the-art methods.

A Game-Theoretical Self-Adaptation Framework for Securing Software-Intensive Systems

Security attacks present unique challenges to the design of self-adaptation mechanism for software-intensive systems due to the adversarial nature of the environment. Game-theoretical approaches have been explored in security to model malicious behaviors and design reliable defense for the system in a mathematically grounded manner. However, modeling the system as a single player, as done in prior works, is insufficient for the system under partial compromise and for the design of fine-grained defensive policies where the rest of the system with autonomy can cooperate to mitigate the impact of attacks. To address such issues, we propose a new self-adaptation framework incorporating Bayesian game theory and model the defender (i.e., the system) at the granularity of components. Under security attacks, the architecture model of the system is automatically translated, by the proposed translation process with designed algorithms, into a multi-player Bayesian game. This representation allows each component to be modelled as an independent player, while security attacks are encoded as variant types for the components. By solving for pure equilibrium (i.e., adaptation response), the system’s optimal defensive strategy is dynamically computed, enhancing system resilience against security attacks by maximizing system utility. We validate the effectiveness of our framework through two sets of experiments using generic benchmark tasks tailored for the security domain. Additionally, we exemplify the practical application of our approach through a real-world implementation in the Secure Water Treatment System to demonstrates the applicability and potency in mitigating security risks.

A Progressive Subnetwork Searching Framework for Dynamic Inference.

Deep neural network (DNN) model compression is a popular and important optimization method for efficient and fast hardware acceleration. However, the compressed model is usually fixed, without the capability to tune the computing complexity (i.e., latency in hardware) on-the-fly, depending on dynamic latency requirements, workloads, and computing hardware resource allocation. To address this challenge, dynamic DNN with run-time adaption of computing structures has been constructed through training with a cross-entropy objective function consisting of multiple subnets sampled from the supernet. Our investigations in this work show that the performance of dynamic inference highly relies on the quality of subnet sampling. To construct a dynamic DNN with multiple high-quality subnets, we propose a progressive subnetwork searching framework, which is embedded with several proposed new techniques, including trainable noise ranking, channel-group sampling, selective fine-tuning, and subnet filtering. Our proposed framework empowers the target dynamic DNN with higher accuracy for all the subnets compared with prior works on both the Canadian Institute for Advanced Research dataset with 10 classes (CIFAR-10) and ImageNet datasets. Specifically, compared with United States-Neural Network (US-NN), our method achieves 0.9% average accuracy gain for Alexnet, 2.5% for ResNet18, 1.1% for Visual Geometry Group (VGG)11, and 0.58% for MobileNetv1, on the ImageNet dataset, respectively. Moreover, to demonstrate run-time tuning of computing latency of dynamic DNN in real computing system, we have deployed our constructed dynamic networks into Nvidia Titan graphics processing unit (GPU) and Intel Xeon central processing unit (CPU), showing great improvement over prior works. The code is available at https://github.com/ASU-ESIC-FAN-Lab/Dynamic-inference.

Few-shot learning based on prototype rectification with a self-attention mechanism

Few-shot learning (FSL) is a challenging problem. Prototype-based methods are simple and effective methods for addressing few-shot learning. Due to the lack of labeled samples, the learned class prototype in the existing prototype-based few-shot learning methods has a great deviation and cannot express the representative and discriminant characteristics of its corresponding class well. To address this problem, in this work we propose few-shot learning based on prototype rectification with a self-attention mechanism(FSL-PRS). To learn more unbiased and discriminative class prototypes, FSL-PRS takes the support set and the query set as a whole and learns task-related features from the features extracted from pretrained backbone networks with a self-attention mechanism. Then, the learned task-related features are utilized to compute the original class prototypes and predict a pseudo label and confidence for each query sample. The query samples with high confidence are incorporated into the support set to rectify the class prototypes. We hope that the learned class prototype can better highlight the class significance. Therefore, a class significance learning module is designed for making the learned class prototypes more discriminative. Different from prior works, we take the support set and the query set as a whole to learn task-related features with a self-attention mechanism, which not only alleviates the negative effect of distribution differences between the support set and query set but also fuses global context information to enhance features for FSL. We conduct comprehensive experiments on four benchmark datasets widely adopted in few-shot learning. The experimental results demonstrate that the FSL-PRS achieves state-of-the-art performance, which validates its effectiveness.

Incremental Learning, Incremental Backdoor Threats

Class incremental learning from a pre-trained DNN model is gaining lots of popularity. Unfortunately, the pre-trained model also introduces a new attack vector, which enables an adversary to inject a backdoor into it and further compromise the downstream models learned from it. Prior works proposed backdoor attacks against the pre-trained models in the transfer learning scenario. However, they become less effective when the adversary does not have the knowledge of the downstream tasks or new data, which is more practical and considered in this paper. To this end, we design the first latent backdoor attacks against incremental learning. We propose two novel techniques, which can effectively and stealthily embed a backdoor into the pre-trained model. Such backdoor can only be activated when the pre-trained model is extended to a downstream model with incremental learning. It has a very high attack success rate, and is able to bypass existing backdoor detection approaches. Extensive experiments confirm the effectiveness of our attacks over different datasets and incremental learning methods, as well as strong robustness against state-of-the-art backdoor defense mechanisms including <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Neural Cleanse</i> , <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Fine-Pruning</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">STRIP</i> .

Assessment of Social Vulnerabilities of Care and Prognosis in Adult Ocular Melanomas in the US.

Prior works have studied the impact of social determinants on various cancers but there is limited analysis on eye-orbit cancers. Current literature tends to focus on socioeconomic status and race, with sparse analysis of interdisciplinary contributions. We examined social determinants as measured by the Centers for Disease Control and Prevention (CDC) Social Vulnerability Index (SVI), quantifying eye and orbit melanoma disparities across the United States. A retrospective review of 15,157 patients diagnosed with eye-orbit cancers in the Surveillance, Epidemiology, and End Results (SEER) database from 1975 to 2017 was performed, extracting 6139 ocular melanomas. SVI scores were abstracted and matched to SEER patient data, with scores generated by weighted averages per population density of county's census tracts. Primary outcome was months survived, while secondary outcomes were advanced staging, high grading, and primary surgery receipt. With increased total SVI score, indicating more vulnerability, we observed significant decreases of 23.1% in months survival for melanoma histology (p<0.001) and 19.6-39.7% by primary site. Increasing total SVI showed increased odds of higher grading (odds ratio [OR] 1.20, 95% confidence interval [CI] 1.02-1.43) and decreased odds of surgical intervention (OR 0.94, 95% CI 0.92-0.96). Of the four themes, higher magnitude contributions were observed with socioeconomic status (26.0%) and housing transportation (14.4%), while lesser magnitude contributions were observed with minority language status (13.5%) and household composition (9.0%). Increasing social vulnerability, as measured by the CDC SVI and its subscores, displayed significant detrimental trends in prognostic and treatment factors for adult eye-orbit melanoma. Subscores quantified which social determinants contributed most to disparities. This lays groundwork for providers to target the highest-impact social determinant for non-clinical factors in patient care.

Magnetic Performance of Eddy Current Suppressing Structures in Additive Manufacturing

Additively manufactured soft-magnetic components are inherently bulky leading to significant eddy current losses when applied to electrical machines. Prior works have addressed this issue by implementing structures based on the Hilbert space-filling curve which include eddy current suppressing gaps, thereby reducing the fill factor of the soft-magnetic component. The present research aims at investigating a number of space-filling curves in addition to sheets in order to find the optimal eddy current suppressing structure from an electromagnetic point of view. By means of both analysis and finite-element simulation, it was shown that sheets are superior at minimizing eddy current losses while space-filling curves excel at maximizing the fill factor.

Graph-based social relation inference with multi-level conditional attention

Social relation inference intrinsically requires high-level semantic understanding. In order to accurately infer relations of persons in images, one needs not only to understand scenes and objects in images, but also to adaptively attend to important clues. Unlike prior works of classifying social relations using attention on detected objects, we propose a MUlti-level Conditional Attention (MUCA) mechanism for social relation inference, which attends to scenes, objects and human interactions based on each person pair. Then, we develop a transformer-style network to achieve the MUCA mechanism. The novel network named as Graph-based Relation Inference Transformer (i.e., GRIT) consists of two modules, i.e., a Conditional Query Module (CQM) and a Relation Attention Module (RAM). Specifically, we design a graph-based CQM to generate informative relation queries for all person pairs, which fuses local features and global context for each person pair. Moreover, we fully take advantage of transformer-style networks in RAM for multi-level attentions in classifying social relations. To our best knowledge, GRIT is the first for inferring social relations with multi-level conditional attention. GRIT is end-to-end trainable and significantly outperforms existing methods on two benchmark datasets, e.g., with performance improvement of 7.8% on PIPA and 9.6% on PISC.

TSwinPose: Enhanced monocular 3D human pose estimation with JointFlow

Monocular estimation of 3D human poses is challenging due to ambiguity in depths and partial occlusion. Most recent works define this as a 2D-to-3D lifting task, taking 2D key point sequences and using spatial and temporal relationships. However, prior works focus on capturing spatio-temporal correlations but ignore the motion of joints that is needed for continuous estimation. To extend the potential of 2D-to-3D pose estimation, we propose TSwinPose, which learns multi-scale spatio-temporal representations from 2D key point locations and patterns of motion. The input 2D key point sequences are enhanced by JointFlow, which encodes the motion of each human joint. Based on Swin-Transformer, we designed a temporal domain Swin-Unet structure to model multi-scale spatio-temporal relationships of human joints across different temporal windows. The final 3D pose generated by multi-stage representations is consistent temporally and has a higher accuracy. Experiments conducted on three benchmark datasets, Human3.6M, MPI-INF-3DHP, and HumanEva-I, demonstrate that TSwinPose achieves performance that is on par with state-of-the-art methods. Moreover, the introduction of JointFlow as a plug-in extension enhances performance significantly, particularly benefiting long-term 2D-to-3D lifting human pose estimation methods.

Green policymaking in top emitters: assessing the consequences of external conflicts, trade globalization, and mineral resources on sustainable development

ABSTRACT The escalating levels of environmental degradation along with escalating conflicts and incidents of trade disputes among regional powers can exert a detrimental influence on the prospects for achieving sustainable development (SUD). Prior empirical works have tended to overlook the contributions of trade globalization and external conflicts to the SUD. Furthermore, prevailing studies on SUD often rely on various environmental metrics, economic growth measures, or adjusted net savings as proxies, which fall short in measuring the multifaceted essence of the SUD concept. Thus, this research work assesses the heterogeneous effects of trade globalization, external conflicts, and financial development on a comprehensive SUD index based on environmental, economic, and social factors from 1992 to 2019 in the twenty highest CO2 emitters. The results derived from the application of the method of moment quantile regressions (MM-QR) reveal that trade globalization exerts a negative impact on SUD but its negative effect shows a diminishing trend when moving from the 10th to 90th quantile. Financial development likewise impedes SUD, with the degree of hindrance diminishing as quantiles ascend from lower to upper percentiles. External conflicts hinder SUD only at higher quantiles with the negative effect demonstrating an increasing trend. Furthermore, mineral resources and fossil energy pose detrimental effects on SUD, and these adverse effects exhibit a mounting trajectory as one moves from lower to upper quantiles. Finally, the policy implications of these novel outcomes are discussed.

Automated peripancreatic vessel segmentation and labeling based on iterative trunk growth and weakly supervised mechanism

Peripancreatic vessel segmentation and anatomical labeling are pivotal aspects in aiding surgical planning and prognosis for patients with pancreatic tumors. Nevertheless, prevailing techniques often fall short in achieving satisfactory segmentation performance for the peripancreatic vein (PPV), leading to predictions characterized by poor integrity and connectivity. Besides, unsupervised labeling algorithms usually cannot deal with complex anatomical variation while fully supervised methods require a large number of voxel-wise annotations for training, which is very labor-intensive and time-consuming. To address these two problems, we propose an Automated Peripancreatic vEssel Segmentation and lAbeling (APESA) framework, to not only highly improve the segmentation performance for PPV, but also efficiently identify the peripancreatic artery (PPA) branches. There are two core modules in our proposed APESA framework: iterative trunk growth module (ITGM) for vein segmentation and weakly supervised labeling mechanism (WSLM) for artery labeling. The ITGM is composed of a series of iterative submodules, each of which chooses the largest connected component of the previous PPV segmentation as the trunk of a tree structure, seeks for the potential missing branches around the trunk by our designed branch proposal network, and facilitates trunk growth under the connectivity constraint. The WSLM incorporates the rule-based pseudo label generation with less expert participation, an anatomical labeling network to learn the branch distribution voxel by voxel, and adaptive radius-based postprocessing to refine the branch structures of the labeling predictions. Our achieved Dice of 94.01% for PPV segmentation on our collected dataset represents an approximately 10% accuracy improvement compared to state-of-the-art methods. Additionally, we attained a Dice of 97.01% for PPA segmentation and competitive labeling performance for PPA labeling compared to prior works. Our source codes will be publicly available at https://github.com/ZouLiwen-1999/APESA.

Correlating pipe dimensions and success of local heat treatment: Developing nomograms to deduce heat treatment parameters

Local post-weld heat treatment (PWHT) is the only option for heat treating field welded joints. Quite often, the success of local PWHT in alleviating the residual stress and tempering the material within the soak band (SB) is dependent on the ability to achieve the required heat treatment temperature and maintain through-thickness temperature gradient (TTG) within the specified limits at the end of heating cycle, whence soaking begins. Field observations reveal the inadequacy of AWS D 10.10 specified parameters viz. rate of heating (ROH), heat band (HB) and insulation band width in not achieving the required TTG for certain pipe dimensions. Although prior works have attempted to address this issue by widening the HB, the capacity of the heating equipment often pose a limitation. In such cases, reducing ROH is a plausible alternative. With, no such prior studies seen in literature, an exhaustive finite-element analysis simulating the local PWHT on 81, SA106GrC pipe samples (diameter and thickness varied in 9 levels each) was performed, thrice. First as per AWS recommendations, second by halving the code deduced ROH and third by doubling the code deduced HB. The trend of important outcomes (TTG, power source rating and energy consumption) with great significance to the heat treatment industry were also compared and analysed. Two nomograms were developed to serve as a ready reckoner for field heat treater in not only assessing the adequacy of heat treatment parameters but also with possible alternatives in achieving the desired TTG using field-available power source.

Unmanned Aerial Vehicle-Based Structural Health Monitoring and Computer Vision-Aided Procedure for Seismic Safety Measures of Linear Infrastructures.

Computer vision in the structural health monitoring (SHM) field has become popular, especially for processing unmanned aerial vehicle (UAV) data, but still has limitations both in experimental testing and in practical applications. Prior works have focused on UAV challenges and opportunities for the vibration-based SHM of buildings or bridges, but practical and methodological gaps exist specifically for linear infrastructure systems such as pipelines. Since they are critical for the transportation of products and the transmission of energy, a feasibility study of UAV-based SHM for linear infrastructures is essential to ensuring their service continuity through an advanced SHM system. Thus, this study proposes a single UAV for the seismic monitoring and safety assessment of linear infrastructures along with their computer vision-aided procedures. The proposed procedures were implemented in a full-scale shake-table test of a natural gas pipeline assembly. The objectives were to explore the UAV potential for the seismic vibration monitoring of linear infrastructures with the aid of several computer vision algorithms and to investigate the impact of parameter selection for each algorithm on the matching accuracy. The procedure starts by adopting the Maximally Stable Extremal Region (MSER) method to extract covariant regions that remain similar through a certain threshold of image series. The feature of interest is then detected, extracted, and matched using the Speeded-Up Robust Features (SURF) and K-nearest Neighbor (KNN) algorithms. The Maximum Sample Consensus (MSAC) algorithm is applied for model fitting by maximizing the likelihood of the solution. The output of each algorithm is examined for correctness in matching pairs and accuracy, which is a highlight of this procedure, as no studies have ever investigated these properties. The raw data are corrected and scaled to generate displacement data. Finally, a structural safety assessment was performed using several system identification models. These procedures were first validated using an aluminum bar placed on an actuator and tested in three harmonic tests, and then an implementation case study on the pipeline shake-table tests was analyzed. The validation tests show good agreement between the UAV data and reference data. The shake-table test results also generate reasonable seismic performance and assess the pipeline seismic safety, demonstrating the feasibility of the proposed procedure and the prospect of UAV-based SHM for linear infrastructure monitoring.

Evaluating Generalization, Bias, and Fairness in Deep Learning for Metal Surface Defect Detection: A Comparative Study

In recent years, deep learning models have led to improved accuracy in industrial defect detection, often using variants of YOLO (You Only Look Once), due to its high performance at a low cost. However, the generalizability, fairness and bias of their outcomes have not been examined, which may lead to overconfident predictions. Additionally, the complexity added by co-occurring defects, single and multi-class defects, and the effect on training, is not taken into consideration. This study addresses these critical gaps by introducing new methodologies for analyzing dataset complexity and evaluating model fairness. It introduces the novel approach of co-occurrence impact analysis, examining how the co-occurrence of defects in sample images affects performance, and introducing new dimensions to dataset preparation and training. Its aim is to increase model robustness in the face of real-world scenarios where multiple defects often appear together. Our study also innovates in the evaluation of model fairness by adapting the disparate impact ratio (DIR) to consider the true positive rate (TPR) across different groups and modifying the predictive parity difference (PPD) metric to focus on biases present in industrial quality control. Experiments demonstrate by cross-validation that the model trained on combined datasets significantly outperforms others in accuracy without overfitting and results in increased fairness, as validated by our novel fairness metrics. Explainability also provides valuable insights on the effects of different training regimes, notably absent in prior works. This work not only advances the field of deep learning for defect detection but also provides a strategic framework for future advancements, emphasizing the need for balanced datasets and considerations of ethics, fairness, bias and generalizability in the deployment of artificial intelligence in industry.

Emergent invariance and scaling properties in the collective return dynamics of a stock market.

A key metric to determine the performance of a stock in a market is its return over different investment horizons (τ). Several works have observed heavy-tailed behavior in the distributions of returns in different markets, which are observable indicators of underlying complex dynamics. Such prior works study return distributions that are marginalized across the individual stocks in the market, and do not track statistics about the joint distributions of returns conditioned on different stocks, which would be useful for optimizing inter-stock asset allocation strategies. As a step towards this goal, we study emergent phenomena in the distributions of returns as captured by their pairwise correlations. In particular, we consider the pairwise (between stocks i, j) partial correlations of returns with respect to the market mode, ci,j(τ), (thus, correcting for the baseline return behavior of the market), over different time horizons (τ), and discover two novel emergent phenomena: (i) the standardized distributions of the ci,j(τ)'s are observed to be invariant of τ ranging from from 1000min (2.5 days) to 30000min (2.5 months); (ii) the scaling of the standard deviation of ci,j(τ)'s with τ admits good fits to simple model classes such as a power-law τ-λ or stretched exponential function [Formula: see text] (λ, β > 0). Moreover, the parameters governing these fits provide a summary view of market health: for instance, in years marked by unprecedented financial crises-for example 2008 and 2020-values of λ (scaling exponent) are substantially lower. Finally, we demonstrate that the observed emergent behavior cannot be adequately supported by existing generative frameworks such as single- and multi-factor models. We introduce a promising agent-based Vicsek model that closes this gap.