Average Running Time Research Articles

Application of Association Rule Mining Algorithm in Talent Training Assessment in Universities Under Social Sustainability

From the contradiction between the difficulty of employment in enterprises and the difficulty of employment in college students, we can see that the strategies of cultivating talents in colleges and universities are biased. This research combines the talent cultivation mode of colleges and universities with the current situation of social development, constructs a comprehensive quality assessment system for students, and uses hierarchical analysis and expert opinions to determine the weight coefficients of each index. In response to the shortcomings of traditional mining algorithms, an improved APRIORI algorithm based on the bit-array matrix is proposed to improve the mining efficiency by introducing the duplicate terms of identification and variable records in data compression. The experimental results show that the improved algorithm can achieve fast convergence on different data sets. The minimum running time of BM Apriori algorithm in different nodes is 102s; under different supports, the average running time of BM Apriori algorithm is 1.49s, and the efficiency is improved by 81.4%. BM Apriori algorithm is used for data mining, and the association rules between courses and students’ grades obtained show that students’ grades are prone to decline during the freshman to sophomore years. Schools should focus on cultivating students’ good learning habits, cultivating excellent talents, and promoting the sustainable development of society.

Performance map and operating condition optimization of industrial-size SOFCs using a PINN surrogate model

Solid Oxide Fuel Cell (SOFC) is an energy conversion technology featuring high efficiency, power density, durability, and fuel compatibility. In practice, however, SOFCs reach high power and high efficiency at different operating conditions. Hence, the optimal feasible operating condition is not straightforward and requires optimization.To resolve this problem, an accurate pretrained Physics-Informed Neural Network (PINN) model was developed as a surrogate of a 2D multi-physics model. For calibration, a planar SOFC with a 10 × 10 cm2 active area was tested at varied operating conditions. The resulting voltage error of the surrogate model was as low as 0.513%. The average runtime of the PINN model was 0.5 ms per case. Moreover, the PINN model accepts cell performance parameters as input and is therefore highly flexible.The surrogate was thereafter employed to generate performance maps that visualize the steady operating states at each combination of H2 flowrate and operating voltage. The voltages of peak power points and anode-safety boundaries were plotted as functions of H2 flowrate, so that the optimal operating conditions are shown graphically and can be tracked conveniently as the cell degrades.Further studies involving hydrocarbon fuels will be carried out in the future.

A fine-tuned YOLOv5 deep learning approach for real-time house number detection.

Detection of small objects in natural scene images is a complicated problem due to the blur and depth found in the images. Detecting house numbers from the natural scene images in real-time is a computer vision problem. On the other hand, convolutional neural network (CNN) based deep learning methods have been widely used in object detection in recent years. In this study, firstly, a classical CNN-based approach is used to detect house numbers with locations from natural images in real-time. Faster R-CNN, MobileNet, YOLOv4, YOLOv5 and YOLOv7, among the commonly used CNN models, models were applied. However, satisfactory results could not be obtained due to the small size and variable depth of the door plate objects. A new approach using the fine-tuning technique is proposed to improve the performance of CNN-based deep learning models. Experimental evaluations were made on real data from Kayseri province. Classic Faster R-CNN, MobileNet, YOLOv4, YOLOv5 and YOLOv7 methods yield f1 scores of 0.763, 0.677, 0.880, 0.943 and 0.842, respectively. The proposed fine-tuned Faster R-CNN, MobileNet, YOLOv4, YOLOv5, and YOLOv7 approaches achieved f1 scores of 0.845, 0.775, 0.932, 0.972 and 0.889, respectively. Thanks to the proposed fine-tuned approach, the f1 score of all models has increased. Regarding the run time of the methods, classic Faster R-CNN detects 0.603 seconds, while fine-tuned Faster R-CNN detects 0.633 seconds. Classic MobileNet detects 0.046 seconds, while fine-tuned MobileNet detects 0.048 seconds. Classic YOLOv4 and fine-tuned YOLOv4 detect 0.235 and 0.240 seconds, respectively. Classic YOLOv5 and fine-tuned YOLOv5 detect 0.015 seconds, and classic YOLOv7 and fine-tuned YOLOv7 detect objects in 0.009 seconds. While the YOLOv7 model was the fastest running model with an average running time of 0.009 seconds, the proposed fine-tuned YOLOv5 approach achieved the highest performance with an f1 score of 0.972.

Open vehicle routing problem with moving shipments at the cross-docking terminal

In this study, one of the internal operations at the Cross-docking Terminal (CDT), moving shipments (MS) from receiving doors to shipping doors of CDT is integrated with Open Vehicle Routing Problem (OVRP) and the problem is indicated here as OVRPCD-MS. As an additional feature, asymmetric distance between any two customers is assigned by incorporating a characteristic of one-way routes between cities in real-life transportation. The objective is to minimize the total transportation cost which incurs travelling cost between customers, service cost at customer points, service cost at the receiving and shipping doors of CDT, cost of moving shipments inside the CDT and finally the cost of hiring fleets of vehicles. To solve the OVRPCD-MS problem, a Mixed Integer Linear Programming (MILP) model is developed. The programming models are implemented in LINGO (version 18) optimization software. Branch and Bound algorithm is employed to solve ten small-scale instances generated randomly. The applicability of the proposed MILP model is observed. The required fleets of vehicles to be hired and run time to reach the optimal solution are determined. The study revealed that the average run time is exponential for small-scale instances. Thus, it can be concluded that this proposed model can be used for last time planning for small-scale instances. Also, the combinatorial nature of the vehicle routing problem makes OVRPCD-MS as NP-hard. Therefore, this study recommends that heuristic or meta-heuristic methods are more appropriate for the large-scale instances of OVRPCD-MS to reach near-optimum solutions.

A variational level set model based on additive decomposition for segmenting noisy images with intensity inhomogeneity

The variational level set model is widely used in image segmentation. However, it suffers from the limited performance when the images are contaminated by noise and intensity inhomogeneity. To solve this problem, a variational level set model based on additive decomposition is proposed in this paper. First, we decompose the image into three components: structure, intensity inhomogeneity and noise, which are regularized by different metrics. Specially, the structure component is regularized as a piecewise constant function in bounded variation (BV) space; the intensity inhomogeneity is modeled as a smooth function that is regularized by H1 space, and the residual containing noise is measured by L2 norm. Furthermore, a new indirect regularization term for level set function is designed to enhance the accuracy of the segmentation outcomes. And then, alternate direction iteration algorithm combining with gradient descent and ADMM to solve the proposed model. Experiments on both synthetic and real images validate the proposed model. Compared with five state-of-the-art models, the experimental results demonstrate that the proposed model exhibits the significant improvement in both accuracy and efficiency, outperforming all other methods. Quantitative evaluations show the average segmentation coefficients for Jaccard, Dice and Accuracy against ground truth are 0.93, 0.95 and 0.96, respectively, and the average running time is 0.80 seconds, further confirming its superior performance.

Novel randomization and iterative based algorithms for the transactions assignment in blockchain problem.

This study focuses on the load balancing of the transactions in the blockchain. The problem is how to assign these transactions to the blocks. The objective is to guarantee a load balancing of the workload in the time of blocks. The proposed problem is an NP-hard one. To face the hardness of the studied problem, the challenge is to develop algorithms that solve the problem approximately. Finding an approximate solution is a real challenge. In this paper, nine algorithms are proposed. These algorithms are based on the dispatching-rules method, randomization approach, clustering algorithms, and iterative method. The proposed algorithms return approximate solutions in a remarkable time. In addition, in this paper, a novel architecture composed of blocks is proposed. This architecture adds the component "Balancer". This component is responsible to call the best-proposed algorithm and solve the scheduling problem in a polynomial time. In addition, the proposed work helps users to solve the problem of big data concurrency. These algorithms are coded and compared. The performance of these algorithms is tested over three classes of instances. These classes are generated based on uniform distribution. The total number of instances tested is 1350. The average gap, execution time, and the percentage of the best-reached value are used as metrics to measure the performance of the proposed algorithms. Experimental results show the performance of these algorithms and a comparison between them is discussed. The experimental results show that the best algorithm is best-mi-transactions iterative multi-choice with 93.9% in an average running time of 0.003 s.

High-Accuracy Positioning in GNSS-Blocked Areas by Using the MSCKF-Based SF-RTK/IMU/Camera Tight Integration

The integration of global navigation satellite system (GNSS) single-frequency (SF) real-time kinematics (RTKs) and inertial navigation system (INS) has the advantages of low-cost and low-power consumption compared to the multiple-frequency GNSS RTK/INS integration system. However, due to the vulnerability of GNSS signal reception, the application of the GNSS SF-RTK/INS integration is limited in complex environments. To improve the positioning accuracy of SF-RTK/INS integration in GNSS-blocked environments, we present a low-cost tight integration system based on BDS/GPS SF-RTK, a low-cost inertial measurement unit (IMU), and a monocular camera. In such a system, a multi-state constraint Kalman filter (MSCKF) is adopted to integrate the single-frequency pseudo-range, phase-carrier, inertial measurements, and visual data tightly. A wheel robot dataset collected under satellite signal-blocked conditions is used to evaluate its performance in terms of position, attitude, and run time, respectively. Results illustrated that the presented model can provide higher position accuracy compared to those provided by the RTK/INS tight integration system and visual-inertial tight integration system. Moreover, the average running time presents the potential of the presented method in real-time applications.

Key frame extraction method for lecture videos based on spatio-temporal subtitles.

Affected by the Corona Virus Disease 2019 (COVID-19), online lecture videos have witnessed an explosive growth. In the face of massive videos, this paper proposes a method for extracting key frames of lecture videos based on spatio-temporal subtitles, which can efficiently and quickly obtain effective information. Firstly, the spatio-temporal slices of subtitle area of the video sequence are extracted and spliced along the time axis to construct the video spatio-temporal subtitle. Then, the video spatio-temporal subtitle is processed in binarization, and the projection method is used to construct the SSPA curve of the video spatio-temporal subtitle. Finally, a selection method for steady-state key frame is designed, that is, the key frame extraction is realized by combining curve edge detection and subtitle existence threshold, which ensures the robustness of the proposed method. The test results of 8 videos show that the average value of the comprehensive index F1-score of the key frame extracted by the algorithm can reach 0.97, the average precision is 0.97, and the average recall rate is 0.98. It can effectively extract the key frames in lecture videos, and compared with other algorithms, the average running time is reduced to 0.072 of the original, which is helpful to extract video information quickly and accurately.

SRL-TR2: A Safe Reinforcement Learning Based TRajectory TRacker Framework

This paper aims to solve the trajectory tracking control problem for an autonomous vehicle based on reinforcement learning methods. Existing reinforcement learning approaches have found limited successful applications on safety-critical tasks in the real world mainly due to two challenges: 1) sim-to-real transfer; 2) closed-loop stability and safety concern. In this paper, we propose an actor-critic-style framework SRL-TR <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$^2$</tex-math> </inline-formula> , in which the RL-based TRajectory TRackers are trained under the safety constraints and then deployed to a full-size vehicle as the lateral controller. To improve the generalization ability, we adopt a light-weight adapter State and Action Space Alignment (SASA) to establish mapping relations between the simulation and reality. To address the safety concern, we leverage an expert strategy to take over the control when the safety constraints are not satisfied. Hence, we conduct safe explorations during the training process and improve the stability of the policy. The experiments show that our agents can achieve one-shot transfer across simulation scenarios and unseen realistic scenarios, finishing the field tests with average running time less than 10 ms/step and average lateral error less than 0.1 m under the speed ranging from 12 km/h to 18 km/h. A video of the field tests is available at https://youtu.be/pjWcN_fV24g.

Case Study: China Field Test of Dissolvable Collet Demonstrates Time and Cost Savings

_ Over the past few years, dissolvable frac plugs have been effective in addressing unconventional completion challenges by reducing the need for intervention and decreasing overall completion time and cost. However, the majority of dissolvable frac plugs on the market are made up of complex components, which increase their likelihood of failing when run in hole, during stimulation, or not fully dissolving before cleanout. This has resulted in skepticism and challenges with the older generation of dissolvable plugs. To address these issues, Vertechs introduced a new dissolvable collet that can withstand high and low temperatures (100–350°F), has a pressure rating of 10,000 psi, and is less than 2.5 in. in length (Fig. 1). The dissolvable collet combines anchoring and sealing functions into a single component using high-expansion dissolvable metal to create a collet structure with slip segments and a metal seal ring, eliminating the need for elastomers as a sealing material. Compared to traditional frac plug designs, the collet design is simpler and more robust, enabling it to be run and pumped down at higher speed, eliminating issues accompanied with high annular velocities, thereby increasing its running speed in the wellbore. Since it uses less material for manufacturing, it is lighter and easier to dissolve than traditional frac plugs; for instance, the collet only weighs around 1.1 lb, which results in faster cleanout runs, allowing the operators to produce their wells sooner. Many traditional frac plugs weigh around 11–13 lb. Field Test Results An operator in China conducted a field test on a four-well pad in an onshore, southwest shale formation using traditional dissolvable frac plugs in two wells and the new dissolvable collets in the other two. The goal was to speed up completion and minimize coiled tubing operation spending. The results for plug-and-perf and cleanout runs were compared between the two types of plugs, and the new design showed significant improvements. When running conventional frac plugs into the wellbore, the wireline crew adhered to the line speed chart provided by the tool suppliers to minimize the impact of annular velocity on the elastomers. To achieve this, the running speed was maintained within a certain limit. Consequently, the average running time per stage in the vertical section was approximately 50 minutes, while the pumpdown time was about 30 minutes per stage in the horizontal section. In contrast, the dissolvable collets were deployed and pumped down at a higher speed due to their uncomplicated and sturdy structure, leading to an average running time of 32 minutes per stage in the vertical section and 19 minutes per stage in the horizontal section. While it is true that the pumpdown time in the horizontal section decreases as the stages move from the toe to the heel, the average time per stage is used for easy comprehension and visualization. When combined with other aspects of the plug-and-perf process, such as setting the plug, perforating, and pull out of hole, it became apparent that significant time savings were achieved during the running and pumpdown phases in both the vertical and horizontal sections of the well. This resulted in a 15% reduction in time spent per stage (Fig. 2).

PTC-Net: Point-Wise Transformer With Sparse Convolution Network for Place Recognition

In the point-cloud-based place recognition area, the existing hybrid architectures combining both convolutional networks and transformers have shown promising performance. They mainly apply the voxel-wise transformer after the sparse convolution (SPConv). However, they can induce information loss by the sparse voxelization and further result in loss propagation to the transformer, significantly degrading the performance of the network, especially in outdoor scenes with complex geometric structures and multiple small objects. To address this issue, we propose a novel Point-wise Transformer with sparse Convolution (PTC). Specifically, SPConv is applied to the sparsely voxelized point cloud to extract local features, which are then converted to the point-based representation via a feature transformation unit (FTU). As such, our PTC can apply a transformer model based on the point-wise representation rather than on the voxel-wise one. To enhance the ability to capture long-range features and reduce the computational complexity of the transformer, we propose a two-step transformer, each with different grouping strategies. Meanwhile, in both steps, the attention matrix is computed with much fewer points by grouping a single point cloud into different attention domains. The experiments show that the PTC-Net can achieve state-of-the-art (SOTA) performance, with an improvement of 3.6% on average recall@1. Furthermore, to demonstrate the effectiveness of the PTC, we introduce an extremely light-weight version, PTC-Net-L, with only one PTC layer and half initial channel dimensions, also achieving SOTA performance in terms of the average recall rate and running time with only 0.08 M parameters.

Design and operation of a Peucedani Radix weeding device based on YOLOV5 and a parallel manipulator.

To avoid excessive use of herbicides in the weeding operations of Peucedani Radix, a common Chinese herb, a precision seedling avoidance and weeding agricultural robot was designed for the targeted spraying of herbicides. The robot uses YOLOV5 combined with ExG feature segmentation to detect Peucedani Radix and weeds and obtain their corresponding morphological centers. Optimal seedling avoidance and precise herbicide spraying trajectories are generated using a PSO-Bezier algorithm based on the morphological characteristics of Peucedani Radix. Seedling avoidance trajectories and spraying operations are executed using a parallel manipulator with spraying devices. The validation experiments showed that the precision and recall of Peucedani Radix detection were 98.7% and 88.2%, respectively, and the weed segmentation rate could reach 95% when the minimum connected domain was 50. In the actual Peucedani Radix field spraying operation, the success rate of field precision seedling avoidance herbicide spraying was 80.5%, the collision rate between the end actuator of the parallel manipulator and Peucedani Radix was 4%, and the average running time of the parallel manipulator for precision herbicide spraying on a single weed was 2 s. This study can enrich the theoretical basis of targeted weed control and provide reference for similar studies.

Level-set evolution for medical image segmentation with alternating direction method of multipliers

Segmentation via level-set methods plays a significant role in natural and medical image analysis. Recently, the distance regularized level-set evolution (DRLSE) model, which is based on variational level-set formulation has been proven very useful. However, the DRLSE model suffers from difficulties related to its computational cost due to the sign distance function (SDF) and Courant-Friedrichs-Lewy (CFL) condition applied by the gradient descent method. This paper analyzes and tests the well-known fast computing algorithm known as alternating direction method of multipliers (ADMM) for level-set based image segmentation. That is, we compute the DRLSE model by using the ADMM algorithm. The comparison between the classical gradient descent method and the proposed ADMM for two-phase and multi-phase level-set formulation is provided. The numerical experiments on standard public data sets, particularly in the case of medical image segmentation, demonstrate the effectiveness and efficiency of the proposed method. The average segmentation coefficients for Dice and Jaccard against ground truth are 0.97 and 0.92, respectively, and the average running time is 1.70 seconds. 0.0932, 0.993, 0.981, and 0.964 are average estimated values for mean absolute distance (MAD), accuracy, sensitivity, and specificity, respectively. The source code can be found at link https://github.com/WaliSamad/DRLSE-ADMM-Medical-Image-Segmentation.

Autonomous plume Near-Source search assisted by intermittent visible plume information using finite state Machine and YOLOv3-tiny

Since a plume’s visual characteristics have not been fully considered and utilized, the traditional plume near-source search methods lose a large amount of potentially available information, resulting in the low efficiency and reliability of near-source search. This paper proposes a novel plume near-source search approach assisted by the intermittent visible plume information using finite state machine and YOLOv3-tiny for a robot. A finite state machine is used to realize the autonomous behavioral decision-making of the robot according to the obtained visible plume information, olfactory information, and lidar information. In addition, the YOLOv3-tiny network is used to detect the intermittent visible plume information obtained by a depth camera to capture the color and texture of the plume. In order to improve the search reliability of the robot in the olfactory tracking behavior, a whale optimization algorithm is utilized to enable the robot to simulate the hunting behavior and the social mechanism of the whale group and conduct olfactory tracking of the plume. The experimental results show that this proposed plume near-source search approach achieves the success rates of 91.67% and 83.33% in windless and windy environments respectively, and the average total running time is 113.8 s and 159.6 s in the experimental environment with obstacles, respectively.

A Study on a Novel Inverted Ultra-Short Baseline Positioning System and Phase Difference Estimation

Autonomous underwater vehicles (AUVs) are important tools for exploring and studying the ocean. To improve the stealthiness of AUV positioning, a new type of inverted ultra-short baseline (i-USBL) positioning system is proposed in this paper. The system uses a surface GPS buoy as a positioning base station to transmit positioning signals to the AUV, which receives the signals through the i-USBL and calculates its own coordinates. In this way, the power consumption of the AUV will be lower, improving its endurance. In addition, azimuth observation is one of the key observation quantities in ultra-short baseline positioning systems. Therefore, this paper derives two phase difference estimation algorithms for azimuth measurement, which are the Least Mean Square (LMS) algorithm and the Discrete Fourier Transform (DFT) algorithm. Simulation results show that when the signal-to-noise ratio (SNR) is less than or equal to 0 dB, the DFT algorithm has higher accuracy; when SNR is greater than or equal to 10 dB, LMS has slightly higher accuracy than DFT; however, DFT is more stable than LMS at all SNRs. When SNR is 20 dB, the maximum azimuth measurement error of LMS and DFT is 0.3301° and 0.2204°, respectively, which can meet the positioning accuracy requirements. The average running times of LMS and DFT are 0.085 s and 0.011 s, respectively, and LMS has a faster running speed. In summary, the LMS algorithm can be used for azimuth observation when the SNR is good, and DFT can be applied when the SNR is poor.

Structure of a semantic segmentation-based defect detection network for laser cladding infrared images

While selecting the most suitable infrared thermal imaging detection scheme for online inspection during laser cladding processing, this paper designs the RespathU-net semantic segmentation defect detection network for cladding coating defects in infrared images. The network is based on the U-net network framework. It is optimized and improved by redesigning the coding network structure, expanding the network perceptual field, and connecting the paths of residuals, thus enhancing the segmentation effect on the defective areas of the melt coating by addressing the problems that the original network cannot realize the end-to-end output and the poor segmentation effect on the complex objects. The generalization performance test and defect detection experiment of the RespathU-net network were conducted using the Kolektor SDD dataset and the infrared dataset constructed in this paper. The designed network is compared with fully convolutional networks, SegNet, U-net, and DeepLab_V3 in terms of average exchange ratio, similarity coefficient, and running time. The results show that the proposed RespathU-net achieves good multi-size feature recognition, and the effect is much better than other semantic segmentation networks. The via experimental results verify that the actual defect detection accuracy of the designed network is 87.01% .

Automated isocenter optimization approach for treatment planning for gyroscopic radiosurgery.

Radiosurgery is a well-established treatment for various intracranial tumors. In contrast to other established radiosurgery platforms, the new ZAP-X® allows for self-shielding gyroscopic radiosurgery. Here, treatment beams with variable beam-on times are targeted towards a small number of isocenters. The existing planning framework relies on a heuristic based on random selection or manual selection of isocenters, which often leads to a higher plan quality in clinical practice. The purpose of this work is to study an improved approach for radiosurgery treatment planning, which automatically selects the isocenter locations for the treatment of brain tumors and diseases in the head and neck area using the new system ZAP-X® . We propose a new method to automatically obtain the locations of the isocenters, which are essential in gyroscopic radiosurgery treatment planning. First, an optimal treatment plan is created based on a randomly selected nonisocentric candidate beam set. The intersections of the resulting subset of weighted beams are then clustered to find isocenters. This approach is compared to sphere-packing, random selection, and selection by an expert planner for generating isocenters. We retrospectively evaluate plan quality on 10 acoustic neuroma cases. Isocenters acquired by the method of clustering result in clinically viable plans for all 10 test cases. When using the same number of isocenters, the clustering approach improves coverage on average by 31 percentage points compared to random selection, 15 percentage points compared to sphere packing and 2 percentage points compared to the coverage achieved with the expert selected isocenters. The automatic determination of location and number of isocenters leads, on average, to a coverage of 97 ± 3% with a conformity index of 1.22 ± 0.22, while using 2.46 ± 3.60 fewer isocenters than manually selected. In terms of algorithm performance, all plans were calculated in less than 2 min with an average runtime of 75 ± 25s. This study demonstrates the feasibility of an automatic isocenter selection by clustering in the treatment planning process with the ZAP-X® system. Even in complex cases where the existing approaches fail to produce feasible plans, the clustering method generates plans that are comparable to those produced by expert selected isocenters. Therefore, our approach can help reduce the effort and time required for treatment planning in gyroscopicradiosurgery.

Characterizing browser-based medical imaging AI with serverless edge computing: towards addressing clinical data security constraints.

Artificial intelligence (AI) has been widely introduced to various medical imaging applications ranging from disease visualization to medical decision support. However, data privacy has become an essential concern in clinical practice of deploying the deep learning algorithms through cloud computing. The sensitivity of patient health information (PHI) commonly limits network transfer, installation of bespoke desktop software, and access to computing resources. Serverless edge-computing shed light on privacy preserved model distribution maintaining both high flexibility (as cloud computing) and security (as local deployment). In this paper, we propose a browser-based, cross-platform, and privacy preserved medical imaging AI deployment system working on consumer-level hardware via serverless edge-computing. Briefly we implement this system by deploying a 3D medical image segmentation model for computed tomography (CT) based lung cancer screening. We further curate tradeoffs in model complexity and data size by characterizing the speed, memory usage, and limitations across various operating systems and browsers. Our implementation achieves a deployment with (1) a 3D convolutional neural network (CNN) on CT volumes (256×256×256 resolution), (2) an average runtime of 80 seconds across Firefox v.102.0.1/Chrome v.103.0.5060.114/Microsoft Edge v.103.0.1264.44 and 210 seconds on Safari v.14.1.1, and (3) an average memory usage of 1.5 GB on Microsoft Windows laptops, Linux workstation, and Apple Mac laptops. In conclusion, this work presents a privacy-preserved solution for medical imaging AI applications that minimizes the risk of PHI exposure. We characterize the tools, architectures, and parameters of our framework to facilitate the translation of modern deep learning methods into routine clinical care.

Abstract 4284: Enhancing subclonal reconstruction algorithm for resolving complex tumor phylogenies from multi-sample tumor DNA sequencing

Abstract Cancer is characterized by the ongoing accumulation of somatic mutations, providing selective advantages that may lead to dysregulated cellular proliferation. While the cancer genome at diagnosis has been extensively studied, many cancer types still lack strong prognostic biomarkers. The continuous acquisition and selection for driver mutations in a population of cancer cells acts as a Darwinian process, resulting in clonal expansions of progressively more aberrant and fit phenotypes. Reconstructing tumor evolution allows us to understand key events that drive cancer progression and patterns of mutation co-occurrence within clones. These evolutionary features guide our understanding of fundamental mechanisms that lead to disease lethality. Inferring tumor evolution from DNA sequencing data is becoming part of routine analysis in cancer research. As sequencing costs drop, sequencing multiple tumor samples from a patient becomes routine. These multiple samples can represent different spatial regions of a tumor, longitudinal samples from a single region or a combination of both. This provides an opportunity to study tumor evolution in much greater detail and accuracy than was previously feasible through single-sample datasets. The most widely used methods to reconstruct the subclonal evolution of a tumor utilize stochastic-search algorithms. These approaches iterate through a parameter space to select phylogenetic solutions that maximize the likelihood of observed sequencing data. They are optimized for low complexity cases, where the size and number or subclones are relatively limited. As tumor subclonal structure increases in complexity, the parameter space grows exponentially, and stochastic-search algorithms become computationally intractable. For instance, recent benchmarking studies have revealed that many methods fail to reconstruct clone trees for data with as few as ten subclones. To circumvent current computational limitations, we developed a deterministic algorithm for subclonal reconstruction that leverages fundamental principles of cancer biology to encode heuristics that reduce the solution space to biologically plausible phylogenies. When applied to samples (4-36 tumors; median 16) from 12 patients with metastatic breast cancer, our method reduced the average runtime ten-fold. We were able to delineate the evolutionary history of up to 57 distinct subclones per patient, which is infeasible with most current methods. Benchmarking using methods developed for the SMCHet DREAM challenge on real and simulated datasets further quantifies the accuracy, resolution, and scalability. We have thus presented a novel method for rapid and optimized reconstruction of tumor evolutionary histories. Citation Format: Helena Winata, Daniel Knight, Juber A. Patel, Nicholas K. Wang, Pier Selenica, Stefan E. Eng, Caroline Kostrzewa, Jaron Arbet, Yingjie Zhu, Ronglai Shen, Jorge Reis-Filho, Pedram Razafi, Paul C. Boutros. Enhancing subclonal reconstruction algorithm for resolving complex tumor phylogenies from multi-sample tumor DNA sequencing. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4284.

Infiltration Approach of Green Environmental Protection Education in the View of Sustainable Development

In today’s world, global ecological environment problems are increasingly serious, and human beings have gradually realized the importance of protecting the ecological environment. With the continuous deepening of the concept of sustainable development, the concept of green education has also aroused great attention in the education sector; moreover, its concept has gradually been applied to education and teaching. However, there are many problems in the application of green education. For example, the current green education system is not sound, the importance of green education in schools is not obvious, and there is no effective green teacher training program. In order to improve the current application of green education, this paper proposes to integrate the concept of sustainable development into the teaching of green environmental protection education. Through the construction of a sound green education system and green education classrooms, green ecological ideas would thus be infiltrated into many aspects of education and teaching. Due to the uneven distribution of educational resources in many regions, the quality of education in these regions is also uneven. In order to improve this situation, this paper also constructs an educational resource distribution model. Furthermore, this paper also analyzes the educational resource allocation model by using a differential evolution algorithm. From the experimental results, under the algorithm in this paper, the average student–teacher ratio of each county was 4.93, and the average number of books per student was 38.92. The average running time of teacher resources and book resources allocated by the model was found to be 1.47 s and 1.39 s, respectively. Under the traditional algorithm, the average student–teacher ratio in each county was 5.93, and the average number of books per student in each county was 31.8. The average running time of teacher resources and book resources allocated by the model was 2.36 s and 2.58 s, respectively. It can be seen from the above data that the algorithm in this paper can effectively optimize the allocation rationality of the educational resource allocation model and shorten the running time of resource allocation.