Autonomous Reaction Network Exploration in Homogeneous and Heterogeneous Catalysis

Autonomous computations that rely on automated reaction network elucidation algorithms may pave the way to make computational catalysis on a par with experimental research in the field. Several advantages of this approach are key to catalysis: (i) automation allows one to consider orders of magnitude more structures in a systematic and open-ended fashion than what would be accessible by manual inspection. Eventually, full resolution in terms of structural varieties and conformations as well as with respect to the type and number of potentially important elementary reaction steps (including decomposition reactions that determine turnover numbers) may be achieved. (ii) Fast electronic structure methods with uncertainty quantification warrant high efficiency and reliability in order to not only deliver results quickly, but also to allow for predictive work. (iii) A high degree of autonomy reduces the amount of manual human work, processing errors, and human bias. Although being inherently unbiased, it is still steerable with respect to specific regions of an emerging network and with respect to the addition of new reactant species. This allows for a high fidelity of the formalization of some catalytic process and for surprising in silico discoveries. In this work, we first review the state of the art in computational catalysis to embed autonomous explorations into the general field from which it draws its ingredients. We then elaborate on the specific conceptual issues that arise in the context of autonomous computational procedures, some of which we discuss at an example catalytic system. Graphical Abstract

A Machine Learning Method for Detection of Surface Defects on Ceramic Tiles Using Convolutional Neural Networks

We propose a simple but effective convolutional neural network to learn the similarities between closely related raw pixel images for feature representation extraction and classification through the initialization of convolutional kernels from learned filter kernels of the network. The binary-class classification of sigmoid and discriminative feature vectors are simultaneously learned together contrasting the handcrafted traditional method of feature extractions, which split feature-extraction and classification tasks into two different processes during training. Relying on the high-quality feature representation learned by the network, the classification tasks can be efficiently conducted. We evaluated the classification performance of our proposed method using a collection of tile surface images consisting of cracked surfaces and no-cracked surfaces. We tried to classify the tiny-cracked surfaces from non-crack normal tile demarcations, which could be useful for automated visual inspections that are labor intensive, risky in high altitudes, and time consuming with manual inspection methods. We performed a series of comparisons on the results obtained by varying the optimization, activation functions, and deployment of different data augmentation methods in our network architecture. By doing this, the effectiveness of the presented model for smooth surface defect classification was explored and determined. Through extensive experimentation, we obtained a promising validation accuracy and minimal loss.

Coverage Path Planning with Semantic Segmentation for UAV in PV Plants

Solar energy is one of the most strategic energy sources for the world’s economic development. This has caused the number of solar photovoltaic plants to increase around the world; consequently, they are installed in places where their access and manual inspection are arduous and risky tasks. Recently, the inspection of photovoltaic plants has been conducted with the use of unmanned aerial vehicles (UAV). Although the inspection with UAVs can be completed with a drone operator, where the UAV flight path is purely manual or utilizes a previously generated flight path through a ground control station (GCS). However, the path generated in the GCS has many restrictions that the operator must supply. Due to these restrictions, we present a novel way to develop a flight path automatically with coverage path planning (CPP) methods. Using a DL server to segment the region of interest (RoI) within each of the predefined PV plant images, three CPP methods were also considered and their performances were assessed with metrics. The UAV energy consumption performance in each of the CPP methods was assessed using two different UAVs and standard metrics. Six experiments were performed by varying the CPP width, and the consumption metrics were recorded in each experiment. According to the results, the most effective and efficient methods are the exact cellular decomposition boustrophedon and grid-based wavefront coverage, depending on the CPP width and the area of the PV plant. Finally, a relationship was established between the size of the photovoltaic plant area and the best UAV to perform the inspection with the appropriate CPP width. This could be an important result for low-cost inspection with UAVs, without high-resolution cameras on the UAV board, and in small plants.

Automatic lexical collocate extraction for corpus-based ontology building and refinement

Traditional corpus-based methods rely on manual inspection and extraction of lexical collocates in the study of selection preferences, which is a very costly, labor-intensive, and time-consuming task. Devising automatic methods for lexical collocate extraction becomes necessary to handle this task and the immensity of corpora available. With a view to leveraging the Sketch Engine platform and in-built corpora, we propose a working prototype of a Lexical Collocate Extractor (LeCoExt) command-line tool that mines lexical collocates from all types of verbs according to their syntactic constituents and Collocate Frequency Score (CFS). This might be the first tool that performs comprehensive corpus-based studies of the selection preferences of individual or groups of verbs exploiting the capabilities offered by Sketch Engine. This tool might facilitate the task of extracting rich lexico-semantic knowledge from diverse corpora in a few seconds and at a click away. We test its performance for ontology building and refinement departing from a previous detailed analysis of stealing verbs carried out by Fernández-Martínez & Faber (2020). We show how the proposed tool is used to extract conceptual-cognitive knowledge from the THEFT scenario and implement it into FunGramKB Core Ontology through the creation and modification of theft-related conceptual units.

Towards classification of stative verbs in view of corpus data

The paper presents work in progress on the compilation and automatic annotation of a dataset comprising examples of stative verbs in parallel Bulgarian-Russian corpora with the goal of facilitating the elaboration of a classification of stative verbs in the two languages based on their lexical and semantic properties. We extract stative verbs from the Bulgarian and the Russian WordNets with their assigned conceptual information (frames) from FrameNet. We then assign the set of probable Bulgarian and Russian stative verbs to the verb instances in a parallel Bulgarian-Russian corpus using WordNet correspondences to filter out unlikely stative candidates. Further, manual inspection will ensure high quality of the resource and its application for the purposes of semantic analysis.

Computer Vision-Based Detection for Delayed Fracture of Bolts in Steel Bridges

The delayed fracture of high-strength bolts occurs frequently in the bolt connections of long-span steel bridges. This phenomenon can threaten the safety of structures and even lead to serious accidents in certain cases. However, the manual inspection commonly used in engineering to detect the fractured bolts is time-consuming and inconvenient. Therefore, a computer vision-based inspection approach is proposed in this paper to rapidly and automatically detect the fractured bolts. The proposed approach is realized by a convolutional neural network- (CNN-) based deep learning algorithm, the third version of You Only Look Once (YOLOv3). A challenge for the detector training using YOLOv3 is that only limited amounts of images of the fractured bolts are available in practice. To address this challenge, five data augmentation methods are introduced to produce more labeled images, including brightness transformation, Gaussian blur, flipping, perspective transformation, and scaling. Six YOLOv3 neural networks are trained using six different augmented training sets, and then, the performance of each detector is tested on the same testing set to compare the effectiveness of different augmentation methods. The highest average precision (AP) of the trained detectors is 89.14% when the intersection over union (IOU) threshold is set to 0.5. The practicality and robustness of the proposed method are further demonstrated on images that were never used in the training and testing of the detector. The results demonstrate that the proposed method can quickly and automatically detect the delayed fracture of high-strength bolts.

Research on the Present Situation and Development Trend of Subway Tunnel Inspection Vehicle

For the operational subway tunnel, the manual inspection accounts for the majority in terms of detecting the diseases and damage of tunnel. The accuracy of manual inspection mainly depends on the professional level of the detection personnel, and the whole detection process always is inefficient, which cannot meet the needs of actual tunnels. To address this issue, the intelligent mobile tunnel detection vehicle emerges as the times require. By using advanced technologies such as laser scanning and high-speed camera array, the subway tunnel detection vehicle has achieved the advantages of simple operation, comprehensive function and automatic detection. However, the current subway tunnel detection vehicle mainly realizes the scanning detection of tunnel surface diseases, and the detection of tunnel structural diseases is less involved. Based on the track and tunnel detection requirements, this study analyzes the current situation and existing problems of subway tunnel detection comprehensively, puts forward the development direction of tunnel structure detection, and the application prospect of intelligent detection vehicle in subway tunnel is prospected.

fNIRS-QC: Crowd-Sourced Creation of a Dataset and Machine Learning Model for fNIRS Quality Control

Despite technological advancements in functional Near Infra-Red Spectroscopy (fNIRS) and a rise in the application of the fNIRS in neuroscience experimental designs, the processing of fNIRS data remains characterized by a high number of heterogeneous approaches, implicating the scientific reproducibility and interpretability of the results. For example, a manual inspection is still necessary to assess the quality and subsequent retention of collected fNIRS signals for analysis. Machine Learning (ML) approaches are well-positioned to provide a unique contribution to fNIRS data processing by automating and standardizing methodological approaches for quality control, where ML models can produce objective and reproducible results. However, any successful ML application is grounded in a high-quality dataset of labeled training data, and unfortunately, no such dataset is currently available for fNIRS signals. In this work, we introduce fNIRS-QC, a platform designed for the crowd-sourced creation of a quality control fNIRS dataset. In particular, we (a) composed a dataset of 4385 fNIRS signals; (b) created a web interface to allow multiple users to manually label the signal quality of 510 10 s fNIRS segments. Finally, (c) a subset of the labeled dataset is used to develop a proof-of-concept ML model to automatically assess the quality of fNIRS signals. The developed ML models can serve as a more objective and efficient quality control check that minimizes error from manual inspection and the need for expertise with signal quality control.

When It Pays to Catch a Swarm—Evaluation of the Economic Importance of Remote Honey Bee (Apis mellifera) Colony Swarming Detection

Precision beekeeping, or precision apiculture, focuses on individual beehive remote monitoring using different measurement systems and sensors. Sometimes, there are debates about the necessity for such systems and the real-life benefits of the substitution of bee colony manual inspection by automatic systems. Remote systems offer many advantages, but also have their disadvantages and costs. We evaluated the economic benefits of the remote detection of the bee colonies’ reproductive state of swarming. We propose two economic models for predicting differences in the benefits of catching a swarm depending on its travel distance. Models are tested by comparing the situation in four different countries (Austria, Ethiopia, Indonesia, and Latvia). The economic model is based on financial losses caused by bee colony swarming and considers the effort needed to catch the swarm following a remote swarm detection event. The economic benefit of catching a swarm after a remote precision beekeeping notification is shown to be a function of the distance/time to reach the apiary. The possible technical range is tempting, but we demonstrated that remote sensing is economically limited by the ability to physically reach the apiary and interact in time, or alternatively, inform a person living close by. An advanced economic model additionally includes the swarm catching probability, which decreases based on travel distance/time. Based on exemplary values from the four countries, the economic potential of detecting and informing beekeepers about swarming events is calculated.