We present a dataset for building detection and classification from very high-resolution satellite imagery with the focus on object-level interpretation of individual buildings. It is meant to provide not only flexible test platform object algorithms but also solid basis comparison city morphologies investigation urban planning. In most current open datasets, buildings are treated either as class landcover in form masks or simple objects defined by separate contours (footprints). Our...

A major challenge of AI + Science lies in their inherent incompatibility: today's is primarily based on connectionism, while science depends symbolism. To bridge the two worlds, we propose a framework to seamlessly synergize Kolmogorov-Arnold Networks (KANs) and science. The highlights KANs' usage for three aspects scientific discovery: identifying relevant features, revealing modular structures, discovering symbolic formulas. synergy bidirectional: KAN (incorporating knowledge into KANs),...

Radial basis function neural networks (RBF NNs) are one of the most useful tools in classification sonar targets.Despite many abilities RBF NNs, low accuracy classification, entrapment local minima, and slow convergence rate disadvantages these networks.In order to overcome issues, sine-cosine algorithm (SCA) has been used train NNs this work.To evaluate designed classifier, two benchmark underwater problems were used.Also, an experimental target was developed practically merits RBFbased...

We present REGLO, a novel methodology for repairing pretrained neural networks to satisfy global robustness and individual fairness properties. A network is said be globally robust with respect given input region if only all the points in are locally robust. This notion of also captures as special case. prove that any counterexample property must exhibit corresponding large gradient. For ReLU networks, this result allows us efficiently identify linear regions violate property. By formulating...

The design of building heating, ventilation, and air conditioning (HVAC) system is critically important, as it accounts for around half energy consumption directly affects occupant comfort, productivity, health. Traditional HVAC control methods are typically based on creating explicit physical models thermal dynamics, which often require significant effort to develop difficult achieve sufficient accuracy efficiency runtime scalability field implementations. Recently, deep reinforcement...

Model-based reinforcement learning has been widely studied for controller synthesis in cyber-physical systems (CPSs). In particular, safety-critical CPSs, it is important to formally certify system properties (e.g., safety, stability) under the learned RL controller. However, as existing methods typically conduct formal verification after learned, often difficult obtain any certificate, even many iterations between and verification. To address this challenge, we propose a framework that...

For spot-welding process, rational path planning of weld points can improve productivity for welding robot. In basic ant colony optimization, at the beginning iteration, pheromone concentration is only related to length, but this time, often contains a lot redundant parts. Therefore, in initial iteration almost difficult reflect advantages and disadvantages node, it requires multiple iterations slowly show path. shortcoming, Monte Carlo-based optimization (MC-IACO) proposed solve robots. The...

As the largest terrestrial ecosystem globally, grasslands and their Gross Primary Productivity (GPP) play a critical role in global carbon cycle, influenced by environmental changes human activities. This study classifies into multiple types, uses trend analysis to investigate temporal spatial of GPP for various grassland types from 2010 2020, extracts approximately 940,000 pixel data identify evaluate factors using best prediction model PLS-PM structural equation model. The results indicate...

Sclera segmentation is crucial for developing automatic eye-related medical computer-aided diagnostic systems, as well personal identification and verification, because the sclera contains distinct features. Deep learning-based has achieved significant success compared to traditional methods that rely on hand-crafted features, primarily it can autonomously extract critical output-related features without need consider potential physical constraints. However, achieving accurate using these...

Mobile exploration is a longstanding challenge in robotics, yet current methods primarily focus on active perception instead of interaction, limiting the robot's ability to interact with and fully explore its environment. Existing robotic approaches via interaction are often restricted tabletop scenes, neglecting unique challenges posed by mobile exploration, such as large spaces, complex action diverse object relations. In this work, we introduce 3D relational graph that encodes relations...

Building heating, ventilation, and air conditioning (HVAC) systems account for nearly half of building energy consumption [Formula: see text] total in the US. Their operation is also crucial ensuring physical mental health occupants. Compared with traditional model-based HVAC control methods, recent model-free deep reinforcement learning (DRL) based methods have shown good performance while do not require development detailed costly models. However, these DRL approaches often suffer from...

Recognizing mangrove species is a challenging task in coastal wetland ecological monitoring due to the complex environment, high similarity, and inherent symmetry within structural features of species. Many coexist, exhibiting only subtle differences leaf shape color, which increases risk misclassification. Additionally, mangroves grow intertidal environments with varying light conditions surface reflections, further complicating feature extraction. Small are particularly hard distinguish...

The authors propose models for the solution of fundamental problem option replication subject to discrete trading, round lotting, and nonlinear transaction costs using state-of-the-art methods in deep reinforcement learning (DRL), including <i>Q</i>-learning, <i>Q</i>-learning with Pop-Art, proximal policy optimization (PPO). Each DRL model is trained hedge a whole range strikes, no retraining needed when user changes another strike within range. are general, allowing plug any pricing...

Summary We describe the development and validation of a novel algorithm for field-development optimization problems document field-testing results. Our is founded on recent developments in bound-constrained multiobjective nonsmooth functions which structure objective either cannot be exploited or are nonexistent. Such situations typically arise when computed as result numerical modeling, such reservoir-flow simulation within context planning reservoir management. propose an efficient...

Neural networks have shown great promises in planning, control, and general decision making for learning-enabled cyber-physical systems (LE-CPSs), especially improving performance under complex scenarios. However, it is very challenging to formally analyze the behavior of neural network based planners ensuring system safety, which significantly impedes their applications safety-critical domains such as autonomous driving. In this work, we propose a hierarchical planner that analyzes...

Neural networks have been increasingly applied to control in learning-enabled cyber-physical systems (LE-CPSs) and demonstrated great promises improving system performance efficiency, as well reducing the need for complex physical models. However, lack of safety guarantees such neural network based controllers has significantly impeded their adoption safety-critical CPSs. In this work, we propose a controller adaptation approach that automatically switches among multiple controllers,...

Future autonomous systems will employ sophisticated machine learning techniques for the sensing and perception of surroundings making corresponding decisions planning, control, other actions. They often operate in highly dynamic, uncertain challenging environment, need to meet stringent timing, resource, mission requirements. In particular, it is critical yet very ensure safety these systems, given uncertainties system inputs, constant disturbances on operations, lack analyzability many...

Neural networks (NNs) playing the role of controllers have demonstrated impressive empirical performance on challenging control problems. However, potential adoption NN in real-life applications has been significantly impeded by growing concerns over safety these NN-controlled systems (NNCSs). In this work, we present POLAR-Express, an efficient and precise formal reachability analysis tool for verifying NNCSs. POLAR-Express uses Taylor model (TM) arithmetic to propagate TMs layer-by-layer...

With the wide use of lithium-ion batteries (LIBs), battery production has caused many problems, such as energy consumption and pollutant emissions. Although life-cycle impacts LIBs have been analyzed worldwide, phase not separately studied yet, especially in China. Therefore, this research focuses on builds an energy–environment–economy (3E) evaluation system. Two factories China were selected for applied research. Case 1 annually produces 0.22 GWh lithium iron phosphate (LFP) batteries,...