Researchers at Xi'an Jiaotong University built a closed‑loop AI framework centered on a large language model that generates and evaluates thousands of carbon structures, rapidly discovering ultra‑hard, highly anisotropic and novel carbon allotropes such as C16_3, C12 and C8 within minutes.
ESA’s ESAC team introduced AnomalyMatch, a semi‑supervised active‑learning framework that, with fewer than ten labeled anomalies, processed roughly 100 million Hubble cutouts in just 2–3 days, uncovering 1,339 distinct anomalous astrophysical objects such as merging galaxies, gravitational lenses, and jellyfish galaxies.
In the age of information overload, this article explains the science behind rapid forgetting, introduces Ebbinghaus’s Forgetting Curve, describes how memory consolidates from short‑term to long‑term storage, and outlines evidence‑based strategies such as active learning, spaced repetition, testing effect, and contextual association to build durable knowledge.
This article explains how active learning can dramatically reduce the amount of labeled data required for training deep neural networks by selecting the most informative and representative samples, and provides a complete Python implementation of a hybrid query strategy (DBAL) with ResNet‑18.
This article explores advanced machine learning techniques for optimizing bid shading in real-time advertising auctions, introducing a mixed censorship multi-task learning framework and a cost-effective active learning strategy to accurately predict winning price distributions and overcome sample selection bias.
This article provides a comprehensive overview of user profiling algorithms, covering the evolution from ontology‑based traditional methods to modern deep‑learning approaches, including structured label prediction, representation learning, active learning, and large‑model integration, while discussing challenges, practical applications, and future research directions.
This article provides a comprehensive overview of user portrait algorithms, covering their historical development, ontology‑based traditional approaches, deep‑learning enhancements, representation‑learning techniques such as lookalike, active‑learning driven iteration, and the integration of large‑model world knowledge, while also discussing current challenges and future research directions.
The article draws parallels between expert viewers dissecting the drama "The Legend of Zhen Huan," efficient paper‑reading techniques, and the active‑prediction plus contrast‑learning approach that underpins modern AI model training, highlighting how proactive thinking boosts both personal and machine learning outcomes.
The article explains practical techniques for choosing and constructing fine‑tuning data for large language models, covering data diversity through similarity‑based clustering, semi‑supervised filtering with binary classifiers, and uncertainty‑driven sampling using perplexity or reward models to build an efficient, low‑cost pipeline.
This presentation discusses the challenges of label scarcity and class imbalance in graph‑based risk‑control scenarios and proposes a semantic‑aware active‑learning framework that combines uncertainty, graph structure, prototype diversity, and double‑channel information alignment to improve node classification performance.
This presentation explores the challenges of label scarcity and class imbalance in graph‑based risk‑control scenarios, proposing semantic‑aware active learning and prototype‑driven sampling strategies to improve node classification performance on imbalanced graph datasets.
Learning enriches knowledge but, when pursued passively or excessively, can crowd out personal thought, stifle creativity, and make thinking rigid, so a healthy approach blends active, experience‑based insight, selective classic sources, and continual self‑reflection to turn knowledge into genuine understanding.
The article details the design and implementation of the Stella data annotation SaaS platform at 58.com, covering its background, evolution, modular architecture, annotation capabilities across text, image, audio, and video, AI‑assisted labeling, storage solutions, quality and efficiency management, as well as localization and licensing considerations.
To boost NetEase Cloud Music’s comment handling, the authors combine active‑learning‑driven relabeling, domain‑specific MLM pretraining, contrastive‑learning‑based sample expansion, and multi‑task BERT sharing to raise sentiment‑classification precision and recall above 90 % and double moderation clean‑rate, while employing prompt‑generated story themes, IP‑focused classifiers, and hot‑word aggregation for effective short‑text topic clustering and scalable, theme‑aware distribution.
This article explains how a pool‑based active learning framework that combines uncertainty sampling (using BADGE, ALPS, or virtual adversarial perturbations) with diversity‑driven clustering can dramatically cut labeling costs for Transformer‑based NLP models, and presents experimental results showing VAPAL’s competitive performance and early‑stage advantages.
This article describes the production challenges of using BERT for large‑scale text classification at Zuoyebang, explores lightweight alternatives such as knowledge distillation, pruning and quantization, and details a teacher‑student‑active‑learning pipeline that trains a TextCNN model to match BERT performance while dramatically reducing GPU consumption and improving throughput.
This article describes how a second‑hand e‑commerce company built a fast, modular image‑classification pipeline using small binary classifiers, efficientNet‑b0, and active‑learning‑driven data annotation to detect prohibited items while keeping inference latency under 200 ms and reducing labeling costs dramatically.
The talk explains how to maximize the value of training data by exploring background on model generalization, automatic data augmentation techniques, denoising strategies, active learning for selecting unlabeled samples, and robust data splitting methods, offering practical guidelines for AI practitioners.
Active learning addresses the high cost of labeling data by iteratively selecting the most informative unlabeled samples for annotation, thereby reducing labeling effort while achieving target model performance, and the article explains its fundamentals, relationship to supervised and semi‑supervised learning, common selection strategies, hybrid methods, and evaluation metrics.
Meituan’s NLP team leverages few‑shot learning—using data‑augmentation, semi‑supervised, ensemble/self‑training, and domain‑adaptation techniques—to cut annotation costs, achieving 1–2 percentage‑point accuracy gains on internal benchmarks and deploying high‑performing models for tasks such as topic classification, fake‑review detection, and sentiment analysis, while planning broader platform and model extensions.
This article presents a comprehensive study on extracting semantic tags from 58.com voice data, detailing the use of active learning to address cold‑start problems, comparing keyword matching, XGBoost, TextCNN, CRNN, and an improved Wide&Deep model, and demonstrating significant reductions in labeling effort and superior F1 scores across multiple experiments.
Active Learning is a machine learning approach that reduces labeling costs by iteratively selecting the most informative samples for human annotation, using various query strategies such as uncertainty sampling, query-by-committee, expected model change, and density-weighted methods, applicable to domains like image classification, security risk control, and anomaly detection.
This article explains active learning techniques that let models select valuable training samples, reducing annotation costs and improving performance, and describes business‑specific adaptations, experiments, and results that demonstrate its effectiveness in content‑safety applications.
This article describes a WeChat hotspot mining platform that integrates multiple data sources, builds quality and prediction models, employs advanced clustering and multi‑granular text matching techniques, and uses generative active learning to efficiently discover, predict, and present news hotspots for users.
The article critiques the conventional step‑by‑step coding curriculum and introduces Jeff Olsen’s “prediction‑first” approach, which uses problem‑driven prediction to engage learners, improve conceptual understanding, and foster deeper programming skills across languages like Java, Python, and C.
This article describes Alibaba's AI system for automatically filtering nauseating product images on Taobao, covering challenges such as cold‑start, class imbalance, and diverse visual features, and detailing solutions like semi‑supervised learning, active learning, OHEM‑cascade, attention mechanisms, and the resulting business impact.
The article explains the STEM SOS teaching approach—an interactive, project‑based model that blends concept lectures, hands‑on experiments, student‑led presentations, and multimedia resources—to boost student participation, confidence, and deep understanding of scientific concepts.
This article presents an Active PU Learning framework that combines active learning with two‑step PU semi‑supervised learning to improve cash‑out fraud detection, reducing labeling costs, enhancing model performance, and achieving a three‑fold increase in identified fraudulent transactions compared to traditional unsupervised methods.
This article presents an Active PU Learning framework that combines active learning with two‑step positive‑unlabeled learning to improve cash‑out fraud detection in Alipay’s fifth‑generation risk engine, AlphaRisk, achieving three‑fold identification gains over unsupervised methods while reducing labeling costs.
The article explains the Learning Pyramid theory, details each learning method’s retention rates, and illustrates its practical use through a U.S. classroom transcript, highlighting the benefits of active, multimodal teaching over passive lecture‑based approaches.
