Engineering Sciences
140 items
XGBoost beat NeuralProphet for Budapest metro M4 demand forecasting, and its SHAP explanations matched expert expectations
Using one year of hourly boarding data from Budapest metro line M4 (216 usable days, 05:00–24:00) restricted to a single station and direction (Kálvin tér toward Kelenföld vasútállomás) and a 6-hour forecast window, the study fairly compared inherently interpretable NeuralProphet with a black-box XGBoost interpreted post hoc via SHAP, finding XGBoost more accurate (MAE 91.0753 vs 230.2803, RMSE 144.7094 vs 363.2115, R-squared 0.9600 vs 0.7896) and SHAP temporal-feature insights consistent with NeuralProphet seasonality components and transport expert knowledge.
This scientometric analysis maps research trends in smart and sustainable architecture in educational environments, with an emphasis on green schools and campuses, and lists the bibliometric tools and representative literature it draws on.
Titled "Scientometric Analysis of Research Trends in Smart and Sustainable Architecture in Educational Environments: With an Emphasis on Green Schools and Campuses," the work is, as far as the loaded text shows, a scientometric/bibliometric analysis that gathers and cites literature on smart and sustainable architecture in educational environments, green schools and campuses, spanning green school assessment, sustainable education space design, smart buildings and smart campuses, BIM and IoT, biophilic design, energy efficiency, and net-zero energy buildings, and it cites scientometric and visualization tools such as Bibliometrix, VOSviewer, and CiteSpace along with methodological sources on co-citation and science mapping; however, the loaded text presents only a reference list and contai
PC1D modelling indicates that layer thickness, doping concentration, and operating temperature jointly shape InGaN solar cell efficiency
Using PC1D numerical modelling, the work examines how layer thickness, doping concentration, and operating temperature affect InGaN solar cell performance, adding a simulation-based contribution to the existing line of InGaN photovoltaic optimization studies.
Study builds a footwear generation model with diffusion plus LoRA and turns shoe design from a linear flow into a data-insight-to-dynamic-optimization loop
Targeting problems AIGC exposes in footwear design such as generation homogeneity, low process feasibility, and disconnection from market demand, the study follows a path of theoretical analysis, technology construction, system verification, and method refinement: it reviews application status and bottlenecks through literature analysis and a questionnaire survey, extracts labels with a code detection algorithm and builds a footwear generation model on a diffusion architecture combined with LoRA, integrates core data such as materials, processes, soles, and lasts into a structured and correlated digital asset library, and forms a planning-AI creator-craftsman collaborative group for multi-end review and a marketing and data feedback loop, thereby constructing an intelligent auxiliary footw
Microsoft Research intern builds a machine learning pipeline that gives 30-60 minute space-weather risk warnings for 66,935 U.S. substations, detecting nearly 80% of major events
Developed during a Microsoft Research summer internship, this end-to-end machine learning pipeline uses solar-wind observations from the L1 Lagrange point, forecasts of the AE and Dst indices, physics-informed constraints, local geological conductivity, and grid-infrastructure data to produce location-specific geomagnetically induced current risk estimates 30-60 minutes ahead for 66,935 substations in the continental United States, detecting nearly 80% of major space-weather events over the 2020-2026 evaluation period, with an AE forecast RMSE of 410.2 nT and a Dst forecast RMSE of 7.2 nT, outperforming the Burton equation on 62.2% of high-activity hours.
PatchHolmes lets an agent read all 100 candidate commits at once, lifting patch-retrieval Recall@1 from 32.63% to 59.95%
PatchHolmes is a two-phase patch retrieval system whose first stage fuses BM25 with time decay and Qwen3-Embedding-8B dense retrieval via RRF into a top-100 candidate set, and whose second stage runs a frozen open-weight LLM agent that views all 100 candidates listwise through four tools (list_candidates, read_commit, read_file_diff, submit_answer) and submits a single best commit; on 809 CVEs from GitHubAD it reaches 59.95% Recall@1, 25.34% above the pointwise classifier Favia and 31.40% above IRCoT, adds 27.32% Recall@1 over the retriever's top pick on identical candidates, and transfers unchanged to PatchFinder_top10 to lift Recall@1 from 24.28% to 39.86%.
Six-domain evaluation of GPT-6 Astra as an embodied policy: navigation leads, hybrid control lifts manipulation success, but direct in-hand control and dense-reference locomotion remain unreliable
This work systematically evaluates GPT-6 Astra as an embodied policy across six domains—gripper manipulation, dexterous manipulation, mobile manipulation, navigation, locomotion, and humanoid loco-manipulation—comparing direct control with hybrid control that cooperates with learned policies or whole-body controllers, finding leading navigation results (92% on RxR instruction following, 82% on HM3D object search), hybrid success of 48% on RoboDojo, 50% on DexJoCo, and 38.7% on RoboCasa365, while direct in-hand control and dense-reference locomotion remain unreliable, with substantial token use and inference latency recorded.
When one observation admits several valid actions, CVAE KL regularization and flow-model Lipschitz smoothness decide whether a policy keeps multimodality, while standard robot simulation benchmarks turn out to be nearly unimodal
The work formalizes multimodality in behavioral cloning, proves that latent-variable policies preserve demonstrated modes only if the latent carries action-conditioned information and that excessive posterior-prior regularization suppresses it, shows that action-space generative policies are limited by the Lipschitz constant of the base-to-action map, validates these mechanisms on synthetic multimodal navigation and a real-robot bimodal tissue-grasping task, and uses a GMM modal-clustering diagnostic to find limited conditional multimodality in Push-T, UR3, LIBERO, and Meta-World, where deterministic regression remains competitive.
NavHarness carries maps, search records and house notes into the next conversation, lifting GOAT-Bench s-SR by 18.6 to 34.8 points
NavHarness is a training-free embodied harness that makes memory processing part of the navigation loop: each task opens a fresh multi-round agentic session that inherits prior experience through maps, task records and house notes, while outcome verification and run-end consolidation decide what later sessions inherit; on GOAT-Bench it improves s-SR over context-only independent sessions by 18.6 points with GPT-6 Astra, 22.6 with Opus 5, 34.8 with GPT-4o and 30.3 with Qwen3.8-27B, and with SLAM-estimated poses Astra reaches 83.7 s-SR with 36.9 e-SR on GOAT-Bench and 85.9 s-SR on IR2R-CE.
Page 1 · showing 10