Latest research news
Plain-language news about AI and computer science research. Every post explains one peer-reviewed study in plain English, and links the research it is based on.
- A speech model pre-trained on unlabeled audio improved phoneme classification, speaker recognition, and spoken sentiment classification
In a 2020 paper from researchers at National Taiwan University, a speech model called Mockingjay improved performance on phoneme classification, speaker recognition, and sentiment classification of…
- Matthews correlation coefficient gives a more truthful score than accuracy or F1 score for binary classification tests, a 2020 study concludes
Binary classification is the task of putting each case into one of two groups. In a 2020 paper in BMC Genomics, Davide Chicco and Giuseppe Jurman report that two common scores for this task, accuracy…
- Automated jailbreak prompts often make large language models override safety guardrails in fewer than twenty queries
In a 2025 paper published at the IEEE Conference on Secure and Trustworthy Machine Learning (SaTML), researchers at the University of Pennsylvania describe an automated attack called Prompt Automatic…
- Racial bias in a Twitter hate speech classifier fell after a reweighting step, a 2020 study finds
In a 2020 paper in PLoS ONE, researchers report that a trained language model labeled tweets written in African-American English as racist, sexist, hateful or offensive more often than tweets written…
- AI-enabled marketing strategies are linked to greater adoption of sustainable practices among consumers in Gulf retail, a survey of 410 executives suggests
In a 2025 paper in Administrative Sciences, researchers at Middle East University and German Jordanian University surveyed 410 retail executives in Gulf Cooperation Council countries. They analyzed…
- A text-to-speech system showed no statistically significant difference from human recordings in a sentence-level test
In a 2024 paper in IEEE Transactions on Pattern Analysis and Machine Intelligence, researchers at Peking University, Microsoft Research Asia, Microsoft, and the University of Surrey describe a…
- Whole-slide pathology model trained on 335,645 images outperformed region-of-interest and slide foundation models on clinical tasks without fine-tuning
In a 2025 paper in Nature Medicine, researchers describe a model called TITAN that learns from whole-slide images and the language of pathology reports. A whole-slide image is a digital scan of an…
- Llama-AVSR reports a 0.77 percent word error rate on the LRS3 audio-visual speech benchmark
Researchers at Imperial College London, Meta, and Fondazione Bruno Kessler describe a model called Llama-AVSR that turns speech into text using sound and video of the speaker's lips. In a 2025 paper…
- Robotic learning system can store knowledge from past tasks and reuse it for longer tasks, researchers report
In a 2025 paper in Nature Machine Intelligence, researchers describe a robotic system that keeps learning as tasks arrive one at a time and later reuses that knowledge for longer jobs. The system…
- Optimized prompt tuning can match fine-tuning across model sizes and language tasks
In a 2022 paper in the Annual Meeting of the Association for Computational Linguistics, researchers from Tsinghua University, the Beijing Academy of Artificial Intelligence, and the Shanghai Qi Zhi…
- Frequent AI tool use correlates with weaker critical thinking, mediated by cognitive offloading
In a 2025 study published in Societies, researchers found that people who used AI tools often had lower critical thinking abilities. The link appeared to run through cognitive offloading, the habit…
- Two machine learning classifiers identified offensive language in Urdu tweets with 86 percent accuracy
In a 2025 paper in IEEE Access, researchers from the University of Peshawar and other institutions collected 12,020 Urdu tweets and labeled them with OLID, a benchmark taxonomy for offensive…
- Spoofed and deepfake speech detection held near earlier results in a harder challenge
In a 2021 paper, researchers describe ASVspoof 2021, the fourth edition of a challenge series that tests ways to protect voice-verification software, the kind of software that confirms who someone is…
- A simulated control method handles uneven input limits in multi-player nonlinear systems without a hard-to-find starting rule
Researchers at Bohai University and Southwest University report a control method for a class of continuous-time nonlinear systems with several controllers. In a 2024 paper in IEEE Transactions on…
- Structured skill libraries with constraint checks improve zero-shot task success in LLM agents, study suggests
In a 2026 comparative study, researchers report that a structured skill library with constraint checks showed advantages for large language model agents, programs that choose actions from language…
- A set of natural images reduced one vision model's accuracy by about 90 percent
In a 2021 paper in the proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, researchers introduced two image datasets that reliably make machine vision…
- Challenge results suggest spoof and deepfake speech detectors stayed accurate under new transmission effects but did not generalize well across real rooms and online speech datasets
In a 2023 paper in IEEE/ACM Transactions on Audio Speech and Language Processing, researchers reported results from the ASVspoof 2021 challenge, which compares systems that detect spoofed and…
- Fuzzy topological system shows higher accuracy on eleven regression datasets
In a 2025 paper in IEEE Transactions on Fuzzy Systems, researchers at Liaoning Shihua University report that a method called a hierarchical fuzzy topological system, or HFTS, showed higher accuracy…
- Uneven data across silos weakened federated learning accuracy, and no existing method outperformed all others in every setting
In a 2022 paper, researchers from the National University of Singapore and Shanghai Jiao Tong University tested federated learning under uneven data conditions. Federated learning is a way for…
- Score-CAM beat earlier visual explanation methods on recognition and localization tests
In a 2020 paper, researchers described Score-CAM, a method that explains why a convolutional neural network classifies an image the way it does. Convolutional neural networks are a type of artificial…
- A 0.32-million-parameter adapter outperforms full fine-tuning and prompt learning in multimodal object tracking
In a 2024 paper in Proceedings of the AAAI Conference on Artificial Intelligence, a peer-reviewed conference, researchers at Tianjin University report a small adapter for object tracking. The adapter…
- A transformer model predicts rates of more than 1,000 diseases about as accurately as single-disease models
In a 2025 paper published in Nature, a peer-reviewed journal, researchers describe Delphi-2M, a model based on the GPT architecture, a transformer design that learns by predicting the next item in a…
- Pixel-level interpretability model outperforms Grad-CAM on every tested metric in a COVID-19 chest X-ray study
In a 2025 paper published in Machine Learning and Knowledge Extraction, a peer-reviewed journal, researchers at Université du Québec à Chicoutimi reported that a new pixel-level interpretability…
- A deep-learning model with fewer trainable parameters classified breast ultrasound images more accurately than a standard method, a 2025 study reports
In a 2025 paper, researchers report that a deep-learning model called INLCL classified breast ultrasound images into benign and malignant groups more accurately than a Support Vector Machine, a…
- Survey links AI-driven learning culture to stronger innovation performance at information technology firms in Karachi, Pakistan
In a 2026 paper published in VINE Journal of Information and Knowledge Management Systems, a peer-reviewed journal, researchers surveyed 348 professionals in information technology firms in Karachi…
- ConvNeXt V2 models improve image classification, detection, and segmentation results
In a 2023 paper published in peer-reviewed conference proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, researchers at Korea Advanced Institute of…
- Simulations show an optimization method reduced path interference in systems that use a reconfigurable intelligent surface for sensing and communication
In a 2025 paper in IEEE Journal on Selected Areas in Communications, a peer-reviewed journal, researchers at New York University Abu Dhabi and New York University modeled a base station that…
- Language models often miss relevant information in the middle of a long input
In a 2024 paper published in Transactions of the Association for Computational Linguistics, a peer-reviewed journal, researchers at Stanford University and the University of California, Berkeley…
- Transformer-based deep learning models predicted drug response with 91.2 percent accuracy
In a 2025 paper published in Drug Research, a peer-reviewed journal, researchers report that Transformer-based deep learning models predicted drug response with 91.2 percent accuracy. Deep learning…
- In a 2023 study, BERT achieved a 96% F1 score at detecting online hate speech
In a 2023 paper published in Applied Artificial Intelligence, a peer-reviewed journal, researchers from King Abdulaziz University and the University of Tabuk tested two ways to find hate speech in…
