validation.jsonl

{"id": "f138d848f508eb5b", "messages": [{"role": "system", "content": "You are Gemma4Tunnel, a careful tunnel boring machine and tunnelling engineering assistant. Use engineering terminology precisely, state assumptions, and avoid inventing source claims."}, {"role": "user", "content": "Extract reusable tunnelling/TBM engineering knowledge from this allowed source excerpt. Preserve important parameters, mechanisms, risks, and caveats. Do not add unsupported claims.\n\nSource: Spectral Angle Mapping and AI Methods Applied in Automatic Identification of Placer Deposit Magnetite Using Multispectral Camera Mounted on UAV; DOI: 10.3390/min12020268; license: cc-by-sa."}, {"role": "assistant", "content": "To overcome this, Figure 7 3⃝shows how ground truth pixels are extracted from post SAM segmentation maps. Ground truths are the pixels known to be 100% occupied by magnetite (target) and non-magnetite (noise) pixels. When two (magnetite and non-magnetite) ground truth pixels are extracted from the 6-bands deep spectral images, we end up having a total of 12 ground truths (6 for each material). By the end of the segmentation (Figure 7 2⃝), we end up with a given total number of magnetite and non-magnetite pixels per segmented image across 6 spectral bands as shown in Figure 7 4⃝. Hence, Figure 7 4⃝is not the true structure of the image datasets but rather a sum of all pixels thought to be magnetite and non-magnetite post SAM. Thereafter, the segmentation maps are inserted into AI algorithms together with their target and noise ground truths (Figure 7 5⃝). The AI algorithms then attempt to classify the whole image scene containing the segmentation map based on the pre-defined ground truths for magnetite and non-magnetite pixels. From Figure 7 6⃝, we see the algorithms identifying and reclassifying pixels which may not have been correctly classified/mapped by the SAM in Figure 7 2⃝. At the end of the AI classification procedure (Figure 7 7⃝), we see that the AI models have better/improved classification capabilities. Though not perfect either as shown by the confusion matrix, the classification capabilities are significantly more viable than pre-AI. In short, the post SAM segmentation maps as well as the ground truths help guide, inform and speed up the decision making process of the supervised AI algorithms in classifying magnetite iron sand and non-magnetite sands. References 1. Sinaice, B.B.; Takanohashi, Y.; Owada, N.; Utsuki, S.; Hyongdoo, J.; Bagai, Z.; Shemang, E.; Kawamura, Y. Automatic magnetite identification at Placer deposit using multi-spectral camera mounted on UAV and machine learning. In Proceedings of the 5th International Future Mining Conference 2021—AusIMM 2021, Online, 6–8 December 2021; pp. 33–42, ISBN 978-1-922395-02-3. 2. Gaffey, C.; Bhardwaj, A. Applications of Unmanned Aerial Vehicles in Cryosphere: Latest Advances and Prospects. Remote Sens. 2020, 12, 948. [CrossRef] 3. Erbe, H.-H.; Udd, J.E.; Sasiadek, J.Z. Mining Automation. IFAC Proc. Vol. 2004, 37, 299–304. [CrossRef] 4. Mohajane, M.; Essahlaoui, A.; Oudija, F.; El Hafyani, M.; Cláudia Teodoro, A. Mapping Forest Species in the Central Middle Atlas of Morocco (Azrou Forest) through Remote Sensing Techniques. IJGI 2017, 6, 275. [CrossRef] 5. Fox, N.; Parbhakar-Fox, A.; Moltzen, J.; Feig, S.; Goemann, K.; Huntington, J. Applications of Hyperspectral Mineralogy for Geoenvironmental Characterisation. Miner. Eng. 2017, 107, 63–77. [CrossRef]\n[page 19] Minerals 2022, 12, 268 19 of 19 6. Sinaice, B.B.; Kawamura, Y.; Kim, J.; Okada, N.; Kitahara, I.; Jang, H. Application of Deep Learning Approaches in Igneous Rock Hyperspectral Imaging. In Proceedings of the 28th International Symposium on Mine Planning and Equipment Selection—MPES 2019; Springer Series in Geomechanics and Geoengineering; Topal, E., Ed.; Springer International Publishing: Cham, Switzerland, 2020; pp. 228–235. [CrossRef] 7. van der Meer, F.D.; van der Werff, H.M.A.; van Ruitenbeek, F.J.A.; Hecker, C.A.; Bakker, W.H.; Noomen, M.F.; van der Meijde, M.; Carranza, E.J.M.; de Smeth, J.B.; Woldai, T. Multi- and Hyperspectral Geologic Remote Sensing: A Review. Int. J. Appl. Earth Obs. Geoinf. 2012, 14, 112–128. [CrossRef] 8. Zhang, X.; Li, P. Lithological Mapping from Hyperspectral Data by Improved Use of Spectral Angle Mapper. Int. J. Appl. Earth Obs. Geoinf. 2014, 31, 95–109. [CrossRef] 9. Ganesh, U.K.; Kannan, S.T. Creation of Hyper Spectral Library and Lithological Discrimination of Granite Rocks Using SVCHR -1024: Lab Based Approach. J. Hyperspectral Remote Sens. 2017, 7, 168–177. 10. Sharma, N.; Sharma, R.; Jindal, N. Machine Learning and Deep Learning Applications-A Vision. Glob. Transit. Proc. 2021, 2, 24–28. [CrossRef] 11. Weyermann, J.; Schläpfer, D.; Hueni, A.; Kneubühler, M.; Schaepman, M. Spectral Angle Mapper (SAM) for Anisotropy Class Indexing in Imaging Spectrometry Data. In Proceedings of the Imaging Spectrometry XIV, San Diego, CA, USA, 17 August 2009; p. 74570B. [CrossRef] 12. Hu, H.; Feng, D.-Z.; Chen, Q.-Y. A Novel Dimensionality Reduction Method: Similarity Order Preserving Discriminant Analysis. Signal Process. 2021, 182, 107933. [CrossRef] 13."}], "source_title": "Spectral Angle Mapping and AI Methods Applied in Automatic Identification of Placer Deposit Magnetite Using Multispectral Camera Mounted on UAV", "source_doi": "10.3390/min12020268", "source_url": "https://www.mdpi.com/2075-163X/12/2/268/pdf?version=1645347690", "source_license": "cc-by-sa", "zotero_item_key": "VPIE7856"}
{"id": "a544e9e31a4bc049", "messages": [{"role": "system", "content": "You are Gemma4Tunnel, a careful tunnel boring machine and tunnelling engineering assistant. Use engineering terminology precisely, state assumptions, and avoid inventing source claims."}, {"role": "user", "content": "Extract reusable tunnelling/TBM engineering knowledge from this allowed source excerpt. Preserve important parameters, mechanisms, risks, and caveats. Do not add unsupported claims.\n\nSource: Reinforcement learning based process optimization and strategy development in conventional tunneling; DOI: 10.1016/j.autcon.2021.103701; license: cc-by."}, {"role": "assistant", "content": "[page 1] Automation in Construction 127 (2021) 103701 Available online 17 April 2021 0926-5805/© 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Reinforcement learning based process optimization and strategy development in conventional tunneling Georg H. Erharter a,*, Tom F. Hansen b, Zhongqiang Liu b, Thomas Marcher a a Graz University of Technology, Institute for Rock Mechanics and Tunnelling, Rechbauerstraße 12, Graz, Austria b Norwegian Geotechnical Institute, Oslo, Norway A R T I C L E I N F O Keywords: Conventional tunneling Reinforcement learning Tunnel excavation strategy Machine learning Excavation sequences A B S T R A C T Reinforcement learning (RL) - a branch of machine learning - refers to the process of an agent learning to achieve a certain goal by interaction with its environment. The process of conventional tunneling shows many similar­ ities, where a geotechnician (agent) tries to achieve a breakthrough (goal) by excavating the rockmass (envi­ ronment) in an optimum way. In this paper we present a novel RL based framework for strategy development for conventional tunneling. We developed a virtual environment with the goal of a tunnel breakthrough and with a deep Q-network as the agent’s architecture. It can choose from different excavation sequences to reach that goal and learns to do so in an economical and safe way by getting feedback from a specially designed reward system. Result analyses show that the optimal policies have great similarities to current practices of sequential tunneling and the framework has the potential to discover new tunneling strategies. 1. Introduction Digitalization in tunneling is an ongoing process including topics like Artificial Intelligence (AI) technology [1], Building Information Modelling [2] or embedded strain measurements in shotcrete lining [3]. Machine Learning (ML) applications in particular have so far mostly focused on classification tasks based on supervised ML: e.g. [4–6] apply artificial neural networks (ANN) for rockmass behavior classification of tunnel boring machine (TBM) data; [7] use supervised learning methods for automatic work progress identification in NATM (New Austrian Tunneling Method) tunneling; see [8–11] for state-of-the-art reviews on this topic. Applications of unsupervised ML are fewer in number and often related to the search for less dependency on labelled datasets or more objectiveness (e.g. [12,13]). By the time of this writing reinforcement learning (RL) is mostly a matter of research, but already shows mature problem-solvers for game like scenarios [14,15]. Currently, there is a transition from academia to real-world prototypes, with RL-examples like the optimization of a manufacturing process [16], for real-time steering of hydrocarbon drilling [17,18], in optimization of power grids [19] and control systems in general, for AUVs [20], and in robotics and other autonomous vehi­ cles. To our knowledge, however, there is no published application of RL to geotechnics in general or tunneling in particular. Today’s conventional tunneling – sometimes referred to as “drill and blast tunneling” or “sequential excavation method” – is the classical way of tunnel construction and is the product of more than a century of engineering experience [21]. Great flexibility to adapt to changing ground conditions is one of the main benefits of this type of tunneling, but technical and economic success of the excavation is dependent on the experience of the involved engineers and workers. Albeit experience is undoubtedly valuable, depending on it sometimes goes along with simple repetition of “proven ways” or even negligence of innovation. Furthermore, developments in conventional tunneling often have a strong connection to their nation of origin (e.g. Austria: New Austrian Tunnelling Method [22], Norway: Norwegian Tunneling Method [23], Italy: New Italian Tunneling Method [24,25], etc.) which raises concerns about biased researchers and engineers. The goal of this study is to take a first step in the direction of a conceptual RL-model for optimum decision making in conventional tunneling that is as free as possible of conser­ vatism and national biases. Furthermore, as shown in the development of the RL-agent AlphaGO [15], RL systems have the potential to find new solutions to old problems and thus discover unimagined strategies."}], "source_title": "Reinforcement learning based process optimization and strategy development in conventional tunneling", "source_doi": "10.1016/j.autcon.2021.103701", "source_url": "https://www.sciencedirect.com/science/article/pii/S0926580521001527", "source_license": "cc-by", "zotero_item_key": "RXHUVV3X"}
{"id": "d434be6d4495b1e8", "messages": [{"role": "system", "content": "You are Gemma4Tunnel, a careful tunnel boring machine and tunnelling engineering assistant. Use engineering terminology precisely, state assumptions, and avoid inventing source claims."}, {"role": "user", "content": "Extract reusable tunnelling/TBM engineering knowledge from this allowed source excerpt. Preserve important parameters, mechanisms, risks, and caveats. Do not add unsupported claims.\n\nSource: Detection of Ionospheric Scintillation Based on XGBoost Model Improved by SMOTE-ENN Technique; DOI: 10.3390/rs13132577; license: cc-by-sa."}, {"role": "assistant", "content": "2017, 64, 391–402. [CrossRef] 23. Lin, M.; Zhu, X.; Luo, Y.; Yang, F. Analysis of ionospheric scintillation detection based on machine learning. In Proceedings of the International Conference on Sensing, Measurement and Data Analytics in the Era of Artificial Intelligence, Xi’an, China, 15–17 October 2020; pp. 357–361. 24. Ludwig-Barbosa, V.; Sievert, T.; Carlström, A.; Pettersson, M.I.; Vu, V.T.; Rasch, J. Supervised detection of ionospheric scintillation in low-latitude radio occultation measurements. Remote Sens. 2021, 13, 1690. [CrossRef] 25. Ferre, R.M.; Fuente, A.D.L.; Lohan, E.S. Jammer classification in GNSS bands via machine learning algorithms. Sensors 2019, 19, 5–7. 26. Munin, E.; Blais, A.; Couellan, N. GNSS multipath detection using embedded deep CNN on Intel (R) Neural Compute Stick. In Proceedings of the 33rd International Technical Meeting of the Satellite Division of the Institute of Navigation, Online, 21–25 September 2020; pp. 2018–2029. 27. Suzuki, T.; Kusama, K.; Amano, Y. NLOS multipath detection using convolutional neural network. In Proceedings of the 33rd International Technical Meeting of the Satellite Division of the Institute of Navigation, Online, 21–25 September 2020; pp. 2989–3000. 28. Li, J.; Zhu, X.; Ouyang, M.; Li, W.; Chen, Z.; Dai, Z. Research on multi-peak detection of small delay spoofing signal. IEEE Access 2020, 8, 151777–151787. [CrossRef] 29. Franzese, G.; Linty, N.; Dovis, F. Semi-supervised GNSS scintillations detection based on deepinfomax. Appl. Sci. 2020, 10, 381. [CrossRef] 30. Dogo, E.M.; Nwulu, N.I.; Twala, B.; Aigbavboa, C. Accessing imbalance learning using dynamic selection approach in water quality anomaly detection. Symmetry 2021, 13, 818. [CrossRef] 31. Luo, X.; Lou, Y.; Xiao, Q.; Gu, S.; Chen, B.; Liu, Z. Investigation of ionospheric scintillation effects on BDS precise point positioning at low-latitude regions. GPS Solut. 2018, 22, 1–12. [CrossRef] 32. Marlia, D.; Wu, F.; Ekawati, S.; Anggarani, S.; Ahmed, W.A.; Nofri, E.; Byambakhuu, G. Ionospheric scintillation mapping at low latitude: Over Indonesia. In Proceedings of the International Geoscience and Remote Sensing Symposium, Fort Worth, TX, USA, 23–28 July 2015; pp. 21–24. 33. Vadakke Veettil, S.; Aquino, M.; Marques, H.A.; Moraes, A. Mitigation of ionospheric scintillation effects on GNSS precise point positioning (PPP) at low latitudes. J. Geod. 2020, 94, 1–10. [CrossRef] 34. Jiao, Y.; Morton, Y.; Taylor, S.; Pelgrum, W. High latitude ionosphere scintillation characterization. In Proceedings of the Institute of Navigation International Technical Meeting, San Diego, CA, USA, 28–30 January 2013; pp. 579–584. 35. Merid, A.; Nigussie, M.; Ayele, A. Investigation of the characteristics of wavelike oscillations of post-sunset equatorial ionospheric irregularity by decomposing fluctuating TEC. Adv. Space Res. 2021, 67, 1210–1221. [CrossRef] 36. Guo, R. Statistical Studies of Radio Wave Amplitude and Phase Scintillation in the Ionosphere. Master’s Thesis, Wuhan University, Wuhan, China, 2019. 37. Spogli, L.; Alfonsi, L.; Romano, V.; De Franceschi, G.; Joao Francisco, G.M.; Hirokazu Shimabukuro, M.; Bougard, B.; Aquino, M. Assessing the GNSS scintillation climate over Brazil under increasing solar activity. J. Atmos. Solar-Terr. Phys. 2013, 105–106, 199–206. [CrossRef] 38. Taylor, S.; Morton, Y.; Marcus, R.; Bourne, H.; Pelgrum, W.; Van Dierendonck, A.J. Ionospheric scintillation receivers performances based on high latitude experiments. In Proceedings of the Institute of Navigation Pacific Positioning, Navigation and Timing Meeting, Honolulu, HI, USA, 22–25 April 2013; pp. 743–751. 39. Xu, D.; Morton, Y.; Akos, D.; Walter, T. GPS multi-frequency carrier phase characterization during strong equatorial ionospheric scintillation. In Proceedings of the 28th International Technical Meeting of the Satellite Division of the Institute of Navigation, Tampa, FL, USA, 14–18 September 2015; pp. 3787–3796. 40. Sun, P. Study on Key Techniques of the Satellite Navigation Signals Carrier Tracking in the Presence of Ionospheric Scintillation. Ph.D. Thesis, Graduate School of National University of Defense Technology, Changsha, China, March 2017. 41. Chen, T.; Guestrin, C. XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; pp. 785–794. 42. He, H.; Garcia, E.A. Learning from imbalanced data."}], "source_title": "Detection of Ionospheric Scintillation Based on XGBoost Model Improved by SMOTE-ENN Technique", "source_doi": "10.3390/rs13132577", "source_url": "https://www.mdpi.com/2072-4292/13/13/2577/pdf?version=1625212202", "source_license": "cc-by-sa", "zotero_item_key": "4YEIWMK3"}
{"id": "279329e377b408af", "messages": [{"role": "system", "content": "You are Gemma4Tunnel, a careful tunnel boring machine and tunnelling engineering assistant. Use engineering terminology precisely, state assumptions, and avoid inventing source claims."}, {"role": "user", "content": "Extract reusable tunnelling/TBM engineering knowledge from this allowed source excerpt. Preserve important parameters, mechanisms, risks, and caveats. Do not add unsupported claims.\n\nSource: Two-phase multi-model automatic brain tumour diagnosis system from magnetic resonance images using convolutional neural networks; DOI: 10.1186/s13640-018-0332-4; license: cc-by-sa."}, {"role": "assistant", "content": "Bhaiya, in 2013 International Conference on Communication Systems and Network Technologies. Brain tumor detection using unsupervised learning based neural network (IEEE, Gwalior, 2013), pp. 573–577 25. S. N. Deepa, B. A. Devi, in 2012 International Conference on Computer Communication and Informatics, ICCCI–2012, 10–12 January, 2012. Artificial neural networks design for classification of brain tumour (IEEE, Coimbatore, 2012), pp. 1–6 26. C. Da, H. Zhang, Y. Sang, in Proceedings of the 18th Asia Pacific Symposium on Intelligent and Evolutionary Systems. Brain CT image classification with deep neural networks, vol. 1 (Springer International Publishing, Cham, 2015), pp. 653–662 27. X. W. Gao, R. Hui, Z. Tian, Classification of CT brain images based on deep learning networks. Comput. Methods Prog. Biomed. 138, 49–56 (2017) 28. Y. Xu, Z. Jia, Y. Ai, F. Zhang, M. Lai, E. I. Chang, in 2015 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP, 19–24 April, 2015. Deep convolutional activation features for large scale brain tumor histopathology image classification and segmentation (IEEE, Brisbane, 2015), pp. 947–951 29. M. Havaei, A. Davy, D. Warde-Farley, A. Biard, A. Courville, Y. Bengio, C. Pal, P.-M. Jodoin, H. Larochelle, Brain tumor segmentation with deep neural networks. Med. Image Anal. 35, 18–31 (2017) 30. P. D. Chang, in Brain lesion: glioma, multiple sclerosis, stroke and traumatic brain injuries: Second International Workshop, BrainLes 2016, with the Challenges on BRATS, ISLES and mTOP 2016, Held in Conjunction with MICCAI 2016, Athens, Greece, October 17, 2016, Revised Selected Papers, ed. by A. Crimi, B. Menze, O. Maier, M. Reyes, S. Winzeck, and H. Handels. Fully convolutional deep residual neural networks for brain tumor segmentation (Springer, Cham, 2016), pp. 108–118 31. A. Casamitjana, S. Puch, A. Aduriz, V. Vilaplana, in Brainlesion: glioma, multiple sclerosis, stroke and traumatic brain injuries: Second International Workshop, BrainLes 2016, with the Challenges on BRATS, ISLES and mTOP 2016, Held in Conjunction with MICCAI 2016, Athens, Greece, October 17, 2016, Revised Selected Papers, ed. by A. Crimi, B. Menze, O. Maier, M. Reyes, S. Winzeck, and H. Handels. 3d convolutional neural networks for brain tumor segmentation: a comparison of multi-resolution architectures (Springer, Cham, 2016), pp. 150–161 32. X. Zhao, Y. Wu, G. Song, Z. Li, Y. Fan, Y. Zhang, in Brainlesion: glioma, multiple sclerosis, stroke and traumatic brain injuries: Second International Workshop, BrainLes 2016, with the Challenges on BRATS, ISLES and mTOP 2016, Held in Conjunction with MICCAI 2016, Athens, Greece, October 17, 2016, Revised Selected Papers, ed. by A. Crimi, B. Menze, O. Maier, M. Reyes, S. Winzeck, and H. Handels. Brain tumor segmentation using a fully convolutional neural network with conditional random fields (Springer, Cham, 2016), pp. 75–87 33. S. Pereira, A. Oliveira, V. Alves, C. A. Silva, in 2017 IEEE 5th Portuguese Meeting on Bioengineering (ENBENG). On hierarchical brain tumor segmentation in MRI using fully convolutional neural networks: a preliminary study, (2017), pp. 1–4. https://doi.org/10.1109/ENBENG.2017.7889452 34. S. Pereira, A. Pinto, V. Alves, C. A. Silva, Brain tumor segmentation using convolutional neural networks in MRI images. IEEE Transactions on Medical Imaging. 35(5), 1240–1251 (2016). https://doi.org/10.1109/TMI. 2016.2538465 35. A. I¸sin, C. Direko˘glu, M. ¸Sah, Review of MRI-based brain tumor image segmentation using deep learning methods. Procedia Comput. Sci. 102, 317–324 (2016). 12th International Conference on Application of Fuzzy Systems and Soft Computing, ICAFS 2016, 29-30 August 2016, Vienna, Austria 36. R. C. Gonzalez, R. E. Woods, Digital Image Processing, 3rd edn. (Prentice-Hall, Inc., Upper Saddle River, 2006) 37. Y. Pan, W. Huang, Z. Lin, W. Zhu, J. Zhou, J. Wong, Z. Ding, Brain tumor grading based on neural networks and convolutional neural networks. 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Milan, Italy, 25–29 August, EMBC, 699–702 (2015) 38. Y. B. Ian Goodfellow, A. Courville, Deep learning. Book in preparation for MIT Press (2016). http://www.deeplearningbook.org. Accessed 14 Sept 2018 39. T. G. Dietterich, G. Bakiri, Solving multiclass learning problems via error-correcting output codes. J. Artif. Intell. Res. 2, 263–286 (1995) 40. T. Windeatt, G."}], "source_title": "Two-phase multi-model automatic brain tumour diagnosis system from magnetic resonance images using convolutional neural networks", "source_doi": "10.1186/s13640-018-0332-4", "source_url": "https://jivp-eurasipjournals.springeropen.com/track/pdf/10.1186/s13640-018-0332-4", "source_license": "cc-by-sa", "zotero_item_key": "2ZZSU54R"}
{"id": "a6787269868ba1b4", "messages": [{"role": "system", "content": "You are Gemma4Tunnel, a careful tunnel boring machine and tunnelling engineering assistant. Use engineering terminology precisely, state assumptions, and avoid inventing source claims."}, {"role": "user", "content": "Extract reusable tunnelling/TBM engineering knowledge from this allowed source excerpt. Preserve important parameters, mechanisms, risks, and caveats. Do not add unsupported claims.\n\nSource: Detection of Ionospheric Scintillation Based on XGBoost Model Improved by SMOTE-ENN Technique; DOI: 10.3390/rs13132577; license: cc-by-sa."}, {"role": "assistant", "content": "All of the trained models were tested on the same dataset mentioned in Table 1. According to the experimental results for the three detection algorithms and that im- proved by the SMOTE-ENN resampling technique based on these training subsets, Figure 9 shows the corresponding trends of each evaluation indicator with sample ratios of 1:1:1 to 7:7:1. Table 6 presents the improvement ratios in detail, based on a comparison between the detection results obtained with training on the resampled datasets and raw datasets.\n[page 15] Remote Sens. 2021, 13, 2577 15 of 22 86 87 88 89 90 91 92 Accuracy (%) Decision tree Raw data SMOTE-ENN Random forest Raw data SMOTE-ENN XGBoost Raw data SMOTE-ENN 84 86 88 90 92 Recall on class 2 (%) 1:1:1 2:2:1 3:3:1 4:4:1 5:5:1 6:6:1 7:7:1 89 90 91 92 93 F-score on class 2 (%) 1:1:1 2:2:1 3:3:1 4:4:1 5:5:1 6:6:1 7:7:1 Training set ratio on class 0/1/2 1:1:1 2:2:1 3:3:1 4:4:1 5:5:1 6:6:1 7:7:1 2.78% growth 2.44% growth 3.56% growth 6.39% decline 2.82% decline 6.37% growth Figure 9. Testing results for detection algorithms regarding overall accuracy (top row), recall value for class 2 (middle row) and F1 score for class 2 (bottom row), with different ratios of classes 0/1/2 for the training sets. From left to right: testing results with decision tree (left column); testing results with random forest (middle column); testing results with XGBoost (right column). These seven training datasets with 100,000 samples were randomly selected from the overall training dataset listed in Table 1, according to different sample ratios of classes 0/1/2. The testing dataset for each detection model trained on the above training dataset is the same as that mentioned in Table 1. Table 6. Improvement ratios for evaluation indicators with different training datasets with different detection algorithms trained on SMOTE-ENN resampled data, compared with raw training data. Training Set 1 2 3 4 5 6 7 Accuracy (%) Decision tree 2.68 2.18 1.63 0.97 1.41 1.76 1.99 Random forest 1.87 1.49 1.66 1.41 1.72 1.86 1.77 XGBoost 1.35 2.36 3.56 2.11 2.69 2.4 2.44 Recall for class 2 (%) Decision tree 1.92 1.27 3.81 2.62 1.84 2.9 3.57 Random forest 1.2 2.76 3.81 3.9 4.98 5.8 5.4 XGBoost 0.19 3.28 3.18 4.67 5.69 4.34 6.37 F1 score for class 2 (%) Decision tree 1.18 0.56 1.03 0.56 0.21 1.11 2.1 Random forest 1.01 1.74 2.16 2.31 2.9 3.18 2.73 XGBoost 0.23 1.49 1.41 1.99 2.5 1.64 2.78 4.5. Analysis on Imbalanced Testing Datasets We also investigated the detection performance of the proposed method for strong scintillation events with different degrees of imbalance in the testing datasets. Table 7 lists the training datasets and testing datasets with different ratios of the three classes. The same 100,000 samples randomly selected in the overall training dataset were used in each model training process, while seven groups of testing datasets were designed with different ratios of classes, roughly ranging from 1:1:1 to 7:7:1. For each testing dataset with 31,779 samples, the dataset consisted of nine segments of one-hour observations consecutively, with 3531 samples included in each hour’s data. Additionally, several such\n[page 16] Remote Sens. 2021, 13, 2577 16 of 22 segments could be included in several groups of testing sets, but each group of the testing set was not involved in the training dataset. Table 7. The distribution information for testing dataset imbalanced with respect to three classes. Dataset Scintillation Intensity Total Ratio Weak (Class 0) Moderate (Class 1) Strong (Class 2) Training set Raw data 46,610 42,514 10,876 100,000 4.05:3.54:1 SMOTE-ENN 43,201 41,468 45,976 130,645 1.04:1:1.11 Testing set 1 11,945 9264 10,570 31,779 1.13:0.87:1 2 11,983 14,290 5506 31,779 2.18:2.60:1 3 13,984 13,052 4743 31,779 2.95:2.75:1 4 14,124 14,595 3060 31,779 4.61:4.77:1 5 14,124 15,102 2553 31,779 5.53:5.92:1 6 16,268 13,140 2371 31,779 6.86:5.54:1 7 14,124 15,728 1927 31,779 7.33:8.16:1 Table 8 records the improvement ratios for the evaluation indicators for different testing datasets with different detection algorithms trained on SMOTE-ENN resampled data, compared with raw training data, while Figure 10 shows the corresponding values of accuracy, recall and F1 scores for class 2. Table 8. Improvement ratios for evaluation indicators for different testing datasets with different detection algorithms trained on SMOTE-ENN resampled data, compared with raw training data."}], "source_title": "Detection of Ionospheric Scintillation Based on XGBoost Model Improved by SMOTE-ENN Technique", "source_doi": "10.3390/rs13132577", "source_url": "https://www.mdpi.com/2072-4292/13/13/2577/pdf?version=1625212202", "source_license": "cc-by-sa", "zotero_item_key": "4YEIWMK3"}
{"id": "80dec3186490afef", "messages": [{"role": "system", "content": "You are Gemma4Tunnel, a careful tunnel boring machine and tunnelling engineering assistant. Use engineering terminology precisely, state assumptions, and avoid inventing source claims."}, {"role": "user", "content": "Extract reusable tunnelling/TBM engineering knowledge from this allowed source excerpt. Preserve important parameters, mechanisms, risks, and caveats. Do not add unsupported claims.\n\nSource: Spectral Angle Mapping and AI Methods Applied in Automatic Identification of Placer Deposit Magnetite Using Multispectral Camera Mounted on UAV; DOI: 10.3390/min12020268; license: cc-by-sa."}, {"role": "assistant", "content": "In addition to Trees, Ensemble algorithms were employed in the form of Bagged Trees, Subspace K-nearest neighbour (KNN) and Random Undersampling (RUS) Boosted Tress. Bagging is essentially an approach to layering different algorithms [26]. The Subspace KNN on the other hand relies on a stochastic process that randomly selects a number of components of the given feature vector such that when a test sample is compared to a reference, only the selected features have nonzero variables [26]. Lastly, the RUS Boosted Tress model is very useful in alleviating the problem of class balancing when some classes of training data have far fewer observations than others [27]. The RUS Boosted Tress models are fit and added to the ensemble sequentially such that each subsequent model increasingly attempts to correct misclassified predictions of the preceding model [26,27]. These algorithms were employed as they are able to substantially improve prediction performance of learning algorithms in the presence of non-informative features in data. For this reason, Ensemble algorithms are generally referred to as a combination of multiple classifiers. In summary, both Trees and Ensemble algorithms are highly useful in performing bi- nary and multi-class classifications; consequently, these algorithms were thought of as best fitting for the data structure and research question posed by this study. Post training, the top performing models were examined in terms of classification performances demonstrated at each of the 3 flight elevations based on post SAM and ground truthing procedures. 2.4.2. How Deep Learning Algorithms Operate DL algorithms, are far more sophisticated in their architecture compared to ML algo- rithms [10]. They are complex and multilayered, and are built to allow data to pass through information processing layers (like a human brain neurons) in highly connected ways; the result is a non-linear transformation of data which cannot be described by a mathematical function [10,18,19]. It does however take a large amount of data compared to ML to train DL algorithms in order to attain high accuracy output models [19]. For this study, a one dimensional (1D) Convolution Neural Network (CNN) is employed. This type of CNN was applied as it is known to be highly capable in classifying imagery and waveform data [6], which coincides with the data structure presented in this study. 3. Experimental and Analytical Results 3.1. UAV Drone Field Analytics From the flight history logs at the different UAV flight elevations, 80, 32 and 8 multi- spectral images at 2 m, 10 m and 20 m, respectively were captured (Table 1). The number of images is governed by the flight heights, as well as the 62.7◦UAV field of view. From the number of images captured, the spatial resolution at each elevation can thus be resolved as shown in Figure 4. Given that each spectral image measures 1300 pixels × 1600 pixels, at 2 m, 10 m and 20 m, the spatial resolutions were 34 m2, 84 m2 and 338 m2, respectively, each multiplied to cover the 2700 m2 study area. Hence, it can be said that the higher the flight elevation, the lower the spatial resolution, the opposite being true at lower flight elevations.\n[page 8] Minerals 2022, 12, 268 8 of 19 Table 1. Multispectral UAV drone post flight records from investigations carried out at three different flight elevations. Reprinted with permission from ref. [1]. 2021 AusIMM [1] (p. 37). UAV Drone Flight Elevation Number of Images Captured Flight Time (Minutes: Seconds) Spatial Area Coverage (m2) Battery Power Consumed during Mission (%) 2 m 80 21: 32 34 69 10 m 32 8: 23 84 29 20 m 8 2: 08 338 7 Figure 4. Illustration of the difference spatial resolutions of each image, attained from capturing multispectral images at 2 m, 10 m, and 20 m flight elevations. Moreover, based on Figure 4 and Table 1 post flight logs, it has been demonstrated that greater flight elevations allow for a wider area to be captured at a given time. This disregards the need to operate the UAV drone at close proximities to the subject where larger spatial coverage is required. This is a great advantage as it saves time, uses less power, and requires less computational resources to process the numerous multispectral images. Though Table 1 demonstrates the implications of flying at the three different elevations, it does not communicate outputs pertaining to the actual identification of magnetite. This is a task meant for the SAM analysis and AI. 3.2."}], "source_title": "Spectral Angle Mapping and AI Methods Applied in Automatic Identification of Placer Deposit Magnetite Using Multispectral Camera Mounted on UAV", "source_doi": "10.3390/min12020268", "source_url": "https://www.mdpi.com/2075-163X/12/2/268/pdf?version=1645347690", "source_license": "cc-by-sa", "zotero_item_key": "VPIE7856"}