A5100443469- Anomaly Detection Techniques and Applications
- Adversarial Robustness in Machine Learning
- Domain Adaptation and Few-Shot Learning
- Machine Learning and Data Classification
- Advanced Neural Network Applications
- Job Satisfaction and Organizational Behavior
- Topic Modeling
- Neural Networks and Applications
- Retirement, Disability, and Employment
- Transportation Planning and Optimization
- Digital Marketing and Social Media
- Natural Language Processing Techniques
- Work-Family Balance Challenges
- Imbalanced Data Classification Techniques
- Network Security and Intrusion Detection
- Technology Adoption and User Behaviour
- Gender Diversity and Inequality
- Computational and Text Analysis Methods
- Metaheuristic Optimization Algorithms Research
- Data Stream Mining Techniques
- Knowledge Management and Sharing
- Data-Driven Disease Surveillance
- Psychological Well-being and Life Satisfaction
- Time Series Analysis and Forecasting
- Consumer Behavior in Brand Consumption and Identification
Beijing University of Posts and Telecommunications
2020-2025
Columbia University
2025
University of Glasgow
2021-2025
University of East Anglia
2025
Norwich Research Park
2025
University of Florida
2014-2024
China United Network Communications Group (China)
2023-2024
University of Wisconsin–Madison
2020-2024
Hebei University of Technology
2024
Renji Hospital
2024
We consider the problem of detecting out-of-distribution images in neural networks. propose ODIN, a simple and effective method that does not require any change to pre-trained network. Our is based on observation using temperature scaling adding small perturbations input can separate softmax score distributions between in- images, allowing for more detection. show series experiments ODIN compatible with diverse network architectures datasets. It consistently outperforms baseline approach by...
Determining whether inputs are out-of-distribution (OOD) is an essential building block for safely deploying machine learning models in the open world. However, previous methods relying on softmax confidence score suffer from overconfident posterior distributions OOD data. We propose a unified framework detection that uses energy score. show scores better distinguish in- and samples than traditional approach using scores. Unlike scores, theoretically aligned with probability density of less...
Out-of-distribution (OOD) detection is critical to ensuring the reliability and safety of machine learning systems. For instance, in autonomous driving, we would like driving system issue an alert hand over control humans when it detects unusual scenes or objects that has never seen during training time cannot make a safe decision. The term, OOD detection, first emerged 2017 since then received increasing attention from research community, leading plethora methods developed, ranging...
Out-of-distribution (OOD) detection has received much attention lately due to its practical importance in enhancing the safe deployment of neural networks. One primary challenges is that models often produce highly confident predictions on OOD data, which undermines driving principle model should only be about in-distribution samples. In this work, we propose ReAct--a simple and effective technique for reducing overconfidence data. Our method motivated by novel analysis internal activations...
Breast cancer is the most common invasive in women and second main cause of death females, which can be classified Benign or Malignant. Research prevention on breast have attracted more concern researchers recent years. On other hand, development data mining methods provides an effective way to extract useful information from complex database, some prediction, classification clustering made according extracted information. In this study, explore relationship between attributes so that...
The global trend of increasing workplace age diversity has led to growing research attention the organizational consequences age-diverse workforces. Prior primarily focused on statistical relationship between and performance without empirically probing potential mechanisms underlying this relationship. Adopting an intellectual capital perspective, we posit that affects via human social capital. Furthermore, examine functional age-inclusive management as two contingent factors shaping effects...
Detecting out-of-distribution (OOD) inputs is a central challenge for safely deploying machine learning models in the real world. Existing solutions are mainly driven by small datasets, with low resolution and very few class labels (e.g., CIFAR). As result, OOD detection large-scale image classification tasks remains largely unexplored. In this paper, we bridge critical gap proposing group-based framework, along novel scoring function termed MOS. Our key idea to decompose large semantic...
As one of the classical problems in economic market, newsvendor problem aims to make maximal profit by determining optimal order quantity products. However, previous models assume that selling price a product is predefined constant and only regard as decision variable, which may result an unreasonable investment decision. In this article, new model first proposed, involves both variables. way, reformulated mixed-variable nonlinear programming problem, rather than integer linear...
Abstract The great advantages of e‐learning have been recognized, and efforts made to promote adoption. Despite valuable research achievements from technology adoption, previous studies built their models different perspectives. In this study, offer a comprehensive model on we integrated these (ie, TAM, TPB IDT) included culture as moderator. Based 45 relevant empirical research, study conducted meta‐analysis explore key determinants users’ attitude behavioral intention adopt e‐learning....
The influence of human resource management on innovation has attracted considerable research attention over the last decade. However, existing studies have primarily focused macro-level architecture, limiting our understanding about cross-level origin innovation. Developing an emergence-based framework, we propose that employee-experienced high-involvement work system (HIWS) promotes by eliciting collective interactions for knowledge exchange and aggregation. Further, investigate...
Detecting out-of-distribution (OOD) data has become a critical component in ensuring the safe deployment of machine learning models real world. Existing OOD detection approaches primarily rely on output or feature space for deriving scores, while largely overlooking information from gradient space. In this paper, we present GradNorm, simple and effective approach detecting inputs by utilizing extracted GradNorm directly employs vector norm gradients, backpropagated KL divergence between...
Machine learning models often encounter samples that are diverged from the training distribution. Failure to recognize an out-of-distribution (OOD) sample, and consequently assign sample in-class label significantly compromises reliability of a model. The problem has gained significant attention due its importance for safety deploying in open-world settings. Detecting OOD is challenging intractability modeling all possible unknown distributions. To date, several research domains tackle...
Out-of-distribution (OOD) detection is essential to prevent anomalous inputs from causing a model fail during deployment. While improved OOD methods have emerged, they often rely on the final layer outputs and require full feedforward pass for any given input. In this paper, we propose novel framework, multi-level out-of-distribution (MOOD), which exploits intermediate classifier dynamic efficient inference. We explore establish direct relationship between data complexity optimal exit level,...
Out-of-distribution (OOD) detection has received much attention lately due to its importance in the safe deployment of neural networks. One key challenges is that models lack supervision signals from unknown data, and as a result, can produce overconfident predictions on OOD data. Previous approaches rely real outlier datasets for model regularization, which be costly sometimes infeasible obtain practice. In this paper, we present VOS, novel framework by adaptively synthesizing virtual...
Out-of-distribution (OOD) detection is a critical task for deploying machine learning models in the open world. Distance-based methods have demonstrated promise, where testing samples are detected as OOD if they relatively far away from in-distribution (ID) data. However, prior impose strong distributional assumption of underlying feature space, which may not always hold. In this paper, we explore efficacy non-parametric nearest-neighbor distance detection, has been largely overlooked...
Out-of-distribution (OOD) detection is vital to safety-critical machine learning applications and has thus been extensively studied, with a plethora of methods developed in the literature. However, field currently lacks unified, strictly formulated, comprehensive benchmark, which often results unfair comparisons inconclusive results. From problem setting perspective, OOD closely related neighboring fields including anomaly (AD), open set recognition (OSR), model uncertainty, since for one...
Detecting out-of-distribution inputs is critical for safe deployment of machine learning models in the real world. However, neural networks are known to suffer from overconfidence issue, where they produce abnormally high confidence both in- and inputs. In this work, we show that issue can be mitigated through Logit Normalization (LogitNorm) -- a simple fix cross-entropy loss by enforcing constant vector norm on logits training. Our method motivated analysis logit keeps increasing during...
Building reliable object detectors that can detect out-of-distribution (OOD) objects is critical yet underexplored. One of the key challenges models lack supervision signals from unknown data, producing over-confident predictions on OOD objects. We propose a new unknown-aware detection framework through Spatial-Temporal Unknown Distillation (STUD), which dis-tills videos in wild and meaningfully regularizes model's decision boundary. STUD first identifies candidate proposals spatial...
Abstract Autoregressive language models, which use deep learning to produce human-like texts, have surged in prevalence. Despite advances these concerns arise about their equity across diverse populations. While AI fairness is discussed widely, metrics measure dialogue systems are lacking. This paper presents a framework, rooted deliberative democracy and science communication studies, evaluate human–AI communication. Using it, we conducted an algorithm auditing study examine how GPT-3...
Abstract The Honey Badger Algorithm (HBA) is a new swarm intelligence optimization algorithm by simulating the foraging behavior of honey badgers in nature. To further improve its convergence speed and accuracy, an improved HBA based on density factors with elementary functions mathematical spirals polar coordinate system was proposed. proposes six for attenuation states functions, introduces expressions diameters angles seven (Fibonacci spiral, Butterfly curve, Rose Cycloid, Archimedean...