Queue
628 cases were harvested and are backed by a real paper, but they fail the test the index publishes on. Each one is listed here with the reason, so the exclusion can be checked rather than trusted.
| Paper | Year | Cited | Why it is held |
|---|---|---|---|
| Evaluating machine learning models for supernova gravitational wave signal c... | 2024 | 7 | assisted This is a human-run comparative evaluation of many ML models' classification accuracy on simulated GW data rather than an AI-produced scientific result; the models are the object of study, and no real EOS determination or discovery was made by an AI system. “Our study builds on previous work by examining how different machine learning models, parameters, and data preprocessing techniques impact classificat...” |
| Machine learning approaches in microbiome research: challenges and best prac... | 2023 | 137 | not a result The paper's primary contribution is a set of best-practice recommendations supported by benchmarking existing methods on existing data, not a new result produced by an AI system. “To assist decision-making, we offer a set of recommendations on algorithm selection, pipeline creation and evaluation, stemming from the COST Action M...” |
| The Structure, Design, and Closed-Loop Motion Control of a Differential Driv... | 2017 | 65 | unsupported no abstract available for this DOI |
| Physics-based generative model of curvature sensing peptides; distinguishing... | 2023 | 49 | unsupported no abstract available for this DOI |
| Innovations in movement and behavioural ecology from camera traps: Day range... | 2021 | 43 | assisted The machine learning behavior classifier is one component inside a human-designed statistical pipeline for estimating day range; the paper's contribution is the DR estimation methodology (DRRB), not an AI end-to-end solving of the task. “DRRB (i.e. estimating DR based on the classification of movement behaviours by a machine learning procedure and the ratio between speeds)” |
| Crop Yield Prediction Using Hybrid Machine Learning Approach: A Case Study o... | 2023 | 40 | assisted The abstract reports comparative fit statistics of human-designed ML models on an existing yield-prediction task, making it a methodological model-comparison study rather than an AI-produced result that itself constitutes the paper's contribution. “The performances of the algorithms are com-pared on different fit statistics such as RMSE, MAD, MAPE, etc., using numeric agronomic traits of 518 lent...” |
| Application of Machine Learning to Bending Processes and Material Identifica... | 2021 | 35 | assisted The shallow ANNs are human-designed surrogate predictors trained on human-run FEM simulations and embedded in human-led material-identification and bending-process workflows, so the AI functions as an assisting tool within a pipeline humans ran and completed rather than as the agent that end-to-end solved the problem. “The required data for designing the ANN solutions are collected from numerical simulation using finite element methodology (FEM), which in turn was va...” |
| Knowledge Discovery on Cryptocurrency Exchange Rate Prediction Using Machine... | 2022 | 34 | assisted The abstract reports XGBoost's prediction accuracy on the pre-existing task of cryptocurrency exchange-rate prediction, with the model acting as the predictive stage of a human-designed data-mining/blockchain pipeline rather than delivering an end-to-end capability that itself constitutes the paper's contribution. “The applied machine learning algorithm is XGBoost, which performs the highest prediction output, after accuracy measurement performance.” |
| Sign Language Recognition with Multimodal Sensors and Deep Learning Methods | 2023 | 34 | unsupported no abstract available for this DOI |
| Microbiome Preprocessing Machine Learning Pipeline | 2021 | 31 | assisted ML classifiers serve only as the evaluation instrument inside a human-designed study of preprocessing steps, whose actual contribution is a preprocessing pipeline/tool rather than an AI-produced result. “We computed the contribution of each step to the accuracy as measured by the Area Under Curve (AUC) of the classification.” |
| Bars formed in galaxy merging and their classification with deep learning | 2020 | 29 | assisted The CNN is an image-classification tool inside a human-run pipeline of N-body simulations, and the paper's contribution is the astrophysical finding about bar formation, not the AI's capability itself. “We look to investigate bar formation in galaxy merging using methods from deep learning to analyse our N-body simulations.” |
| Visualization of Neural Network Predictions for Weather Forecasting | 2018 | 28 | not a result The paper's contribution is an interactive visualization tool for inspecting neural network outputs and tuning hyper-parameters, i.e., a framework that enables analysis rather than a primary result the AI produced. “we design a comprehensive and interactive system that allows users to study the output of recurrent neural networks on both the complete training data...” |
| Explaining deep learning of galaxy morphology with saliency mapping | 2022 | 28 | unsupported no abstract available for this DOI |
| Impact of non-normal error distributions on the benchmarking and ranking of... | 2020 | 26 | not a result The paper's contribution is a methodological commentary on benchmarking practice, re-running an existing model (KRR-SLATM-L2) on an existing dataset (QM7b) to critique the MAE metric, with no new problem solved by the AI itself. “This article illustrates the limitations of using the Mean Absolute Error (MAE) for benchmarking” |
| Neural network potential energy surface for the low temperature ring polymer... | 2021 | 25 | assisted The neural network serves as a fitting tool to construct the potential energy surface, but the primary scientific results (rate constants, mechanism analysis) come from QCT and RPMD simulations that the researchers ran on that surface, making the AI one component in a human-driven workflow. “Quasi-classical trajectory (QCT) and ring polymer molecular dynamics (RPMD) studies have been carried on this PES to evaluate the rate constant temper...” |
| A large-scale optical microscopy image dataset of potato tuber for deep lear... | 2020 | 25 | not a result This is a dataset description paper whose contribution is the annotated image collection itself; the CNN and UNet results only serve to validate annotation quality, not to solve a primary problem. “We present a new large-scale three-fold annotated microscopy image dataset, aiming to advance the plant cell biology research” |
| Pre-service Language Teachers’ Task-specific Large Language Model Prompting... | 2025 | 25 | assisted The LLMs served as tools that assisted pre-service teachers in a human-run workflow of lesson plan improvement, and the paper's contribution is a qualitative description of those human prompting practices, not an AI-produced result. “The current study explored how pre-service language teachers interacted with LLMs to assist them in improving lesson plans” |
| Prediction of Poststroke Depression Based on the Outcomes of Machine Learnin... | 2022 | 24 | assisted This is a human-designed and human-run clinical study in which standard machine learning algorithms served as the analytical method on researcher-collected assessment data, producing accuracy metrics that informed the authors' clinical conclusion about PSD predictability rather than an AI system autonomously solving the problem end to end. “PSD was successfully predicted using a support vector machine with a radial basis function kernel function (area under curve (AUC) = 0.711, accuracy =...” |
| Data imputation strategies for process optimization of laser powder bed fusi... | 2023 | 22 | assisted The ML imputation methods (kNN, MICE, GINN) are components investigated within a human-run data-collation and evaluation workflow, with the paper's contribution being the human-performed method comparison and ensemble strategy rather than an AI end-to-end solving of a task. “Various imputation techniques of missing data of the SLM Ti6Al4V dataset, such as the k-nearest neighbor (kNN), multivariate imputation by chained equ...” |
| Study on growth rate of TiO <sub>2</sub>... | 2013 | 20 | assisted The ANN is one modeling component within a human-run workflow (sol-gel experiments, MD simulation, AFM/SEM characterization) and merely fits experimental data to predict film thickness, with the paper's contribution being the film characterization rather than the AI capability itself. “the outputs from experimental data were analyzed by using the ANN in order to investigate the effects of deposition process parameters on the film thi...” |
| Predicting Online Shopping Behavior: Using Machine Learning and Google Analy... | 2024 | 18 | assisted The ML models classify engagement levels as one component, but the paper's stated primary contribution is generating recommendations for human marketers and decision-makers, so the AI output informs humans rather than solving the problem itself. “Primarily, this paper goas is to generate a series of recommendations to help the decision-makers and marketers optimizing the marketing strategies.” |
| Quantifying uncertainty in deep learning approaches to radio galaxy classifi... | 2022 | 16 | unsupported no abstract available for this DOI |
| Prognosis Essay Scoring and Article Relevancy Using Multi-Text Features and... | 2017 | 15 | assisted The paper's primary result is a proposed model's accuracy metrics (Cohen's kappa) on existing benchmark datasets relative to prior approaches, i.e., a prediction component intended to automate/assist human evaluation rather than an AI-produced standalone result. “The model achieves kappa values of 0.80 and 0.83 for the first and second data sets, respectively. Kappa values show that the proposed model has bette...” |
| Land Cover Mapping from Colorized CORONA Archived Greyscale Satellite Data a... | 2021 | 14 | assisted The deep learning component (DeOldify) only colorized the greyscale image as a preprocessing step, while the actual land cover classification result was produced by a human-run pipeline of geometric correction, segmentation, training-sample selection, and an SVM classifier. “The DeOldify Deep learning method embedded in the MyHeritage platform was initially applied to colourise the CORONA image.” |
| Benchmarking deep learning models for surface defect detection: a reproducib... | 2025 | 13 | not a result The paper's contribution is a statistical benchmarking methodology that re-runs existing object detection models on an existing dataset, rather than an AI system producing a primary result. “The proposed methodology provides a reliable and reproducible framework for comparing results across models.” |
| Radio Galaxy Zoo: giant radio galaxy classification using multidomain deep l... | 2021 | assisted The abstract reports a human-run method evaluation, comparing multi-domain vs. single-domain CNN classification accuracy on an existing labeled task, rather than an AI-produced scientific result such as newly discovered giant radio galaxies, so the AI is the object of study inside a human-driven workflow. “the inclusion of multi-resolution survey data results in correction of 39% of the misclassifications seen from equivalent single domain networks for t...” | |
| Rate coefficients of the H + H2O2 → H2 + HO2 reaction on an accurate fundame... | 2018 | 24 | assisted The neural network serves only as the potential energy surface input to human-run dynamics methods (RPMD, QCT, CVT/SCT) that actually computed the rate coefficients, making the AI one component in a larger simulation workflow. “calculated using the ring polymer molecular dynamics (RPMD), quasi-classical trajectory (QCT), and canonical variational transition state theory (CVT)...” |
| Vibrational energies for HFCO using a neural network sum of exponentials pot... | 2016 | 24 | unsupported no abstract available for this DOI |