510 of 510 verified cases.
Farmers need detailed short-to-medium-range, high-resolution local forecasts, but major numerical weather prediction models only provide 6-hourly forecasts at 10-20 km grid length.
Pathological estimation of viable versus necrotic tumor after chemotherapy in osteosarcoma requires laborious analysis of whole slide images.
Training drones to avoid obstacles without the laborious labeled datasets that supervised learning requires.
Forensic investigators need to estimate the postmortem interval (time since death) for cadavers found after unknown time periods, and current methods have substantial limitations.
Robots usually learn locomotion and manipulation separately; unifying both skills on reconfigurable limbs lacked data-driven evidence.
Road defects cause health, safety, and vehicle-damage costs, so a low-cost onboard system was built to detect and classify road quality.
Automatic extraction of indoor spatial information from floor plans was mostly demonstrated only on small, simple, low-resolution buildings.
Determining kinetic parameters for large-scale metabolic models is difficult, hindering integration of omics data and broader adoption of these models.
Transformer models for land use and land cover (LULC) analysis from satellite imagery are too computationally expensive for practical use.
Existing metrics like BLEU and BERTScore fail to capture semantic nuances and domain-specific terminology when evaluating code summaries.
Symbolic regression — finding mathematical equations that fit data — is computationally hard.
Land cover classification is hindered by time-consuming annotation, imagery download, and pre-processing tasks that discourage policy makers from creating and using new land cover maps.
Simulated annealing suffers slow convergence and gets stuck in local search spaces when solving combinatorial optimization problems like Max-Cut, nurse scheduling, and TSP.
Classifying short light curves of low Earth orbit resident space objects to infer object type, attitude and shape.
Finding a healthy neuron in acute brain slices for patch clamp recording is done manually, is slow, and challenges novices.
Automatically recognizing emotions in human speech at the utterance level, needed for applications from mental disease diagnosis to human-robot interaction.
Predicting paddy yield in the Cauvery Delta Zone, Tamil Nadu, is difficult because yields depend on interacting genotype, environmental and management factors; statistical methods were inaccurate.
There was no objective visual method to identify individual animals of species lacking unique body markings, such as brown bears.
Automatic speech recognition performs poorly on dysarthric (impaired) speech, which is unintelligible, irregular, and data-scarce.
Reliably recognizing and mapping windthrow areas in forests from very-high-resolution satellite imagery.
Identifying which proteins bind specific non-coding RNAs with wet-lab high-throughput techniques is expensive and time-consuming, so computational prediction is needed.
Automatically classifying land scenes in high-resolution satellite imagery is difficult due to complex textures, cluttered backgrounds, small objects and large scale variations.
Determining protein secondary structures experimentally via X-ray, cryo-EM or NMR is time-consuming and expensive.
Commercial building HVAC systems consume high energy costs while needing to maintain acceptable indoor air quality.