Search

We explore an attenuation and shape-based identification of euhedral pyrites in high-resolution X-ray Computed Tomography (XCT) data using deep neural networks. To deal with the scarcity of annotated data we generate a complementary training set of synthetic images. To investigate and address the domain gap between the synthetic and XCT data, several deep learning models, with and without domain adaption, are trained and compared. We find that a model trained on a small set of human annotations, while displaying over-fitting, can rival the human annotators. The unsupervised domain adaptation approaches are successful in bridging the domain gap, which significantly improves their performance. A domain-adapted model, trained on a dataset that fuses synthetic and real data, is the overall best-performing model. This highlights the possibility of using synthetic datasets for the application of deep learning in mineralogy., This work has been supported by the Industrial Graduate School Collaborative AI and Robotics funded by the Swedish Knowledge Foundation Dnr:20190128 and in collaboration with the industrial partner Orexplore Technologies.

Zinc (Zn) is an essential element that influences many cellular functions. Depending on bioavailability, Zn can cause both deficiency and toxicity. Zn bioavailability is influenced by water hardness. Therefore, water quality analysis for health-risk assessment should consider both Zn concentration and water hardness. However, exposure media selection for traditional toxicology tests are set to defined hardness levels and do not represent the diverse water chemistry compositions observed in nature. Moreover, these tests commonly use whole organism endpoints, such as survival and reproduction, which require high numbers of test animals and are labor intensive. Gene expression stands out as a promising alternative to provide insight into molecular events that can be used for risk assessment. In this work, we apply machine learning techniques to classify the Zn concentrations and water hardness from Daphnia magna gene expression by using quantitative PCR. A method for gene ranking was explored using techniques from game theory, namely, Shapley values. The results show that standard machine learning classifiers can classify both Zn concentration and water hardness simultaneously, and that Shapley values are a versatile and useful alternative for gene ranking that can provide insight about the importance of individual genes.

Finding a balance between meeting test coverage and minimizing the testing resources is always a challenging task both in software (SW) and hardware (HW) testing. Therefore, employing machine learning (ML) techniques for test optimization purposes has received a great deal of attention. However, utilizing machine learning techniques frequently requires large volumes of valuable data to be trained. Although, the data gathering is hard and also expensive, manual data analysis takes most of the time in order to locate the source of failure once they have been produced in the so-called fault localization. Moreover, by applying ML techniques to historical production test data, relevant and irrelevant features can be found using strength association, such as correlation- and mutual information-based methods. In this paper, we use production data records of 100 units of a 5G radio product containing more than 7000 test results. The obtained results show that insightful information can be found after clustering the test results by their strength association, most linear and monotonic, which would otherwise be challenging to identify by traditional manual data analysis methods.

Q-Networks are used in Reinforcement Learning to model the expected return from every action at a given state. When training Q-Networks, external reward signals are propagated to the previously performed actions leading up to each reward. If many actions are required before experiencing a reward, the reward signal is distributed across all those actions, where some actions may have greater impact on the reward than others. As the number of significant actions between rewards increases, the relative importance of each action decreases. If actions have too small importance, their impact might be over-shadowed by noise in a deep neural network model, potentially causing convergence issues. In this work, we empirically test the limits of increasing the number of actions leading up to a reward in a simple grid-world environment. We show in our experiments that even though the training error surpasses the reward signal attributed to each action, the model is still able to learn a smooth enough value representation.

The Green Area Factor(GYF) is an aggregate norm used as an index to quantify how much eco-efficient surface exists in a given area. Although the GYF is a single number, it expresses several different contributions of natural objects to the ecosystem. It is used as a planning tool to create and manage attractive urban environments ensuring the existence of required green/blue elements. Currently, the GYF model is gaining rapid attraction by different communities. However, calculating the GYF value is challenging as significant amount of manual effort is needed. In this study, we present a novel approach for automatic extraction of the GYF value from aerial imagery using semantic segmentation results. For model training and validation a set of RGB images captured by Drone imaging system is used. Each image is annotated into trees, grass, soil/open surface, building, and road. A modified U-net deep learning architecture is used for the segmentation of various objects by classifying each pixel into one of the semantic classes. From the segmented image we calculate the class-wise fractional area coverages that are used as input into the simplified GYF model called Sundbyberg for calculating the GYF value. Experimental results yield that the deep learning method provides about 92% mean IoU for test image segmentation and corresponding GYF value is 0.34.

Software testing is still heavily dependent on human judgment since a large portion of testing artifacts, such as requirements and test cases are written in a natural text by experts. Identifying and classifying relevant test cases in large test suites is a challenging and also time-consuming task. Moreover, to optimize the testing process test cases should be distinguished based on their properties, such as their dependencies and similarities. Knowing the mentioned properties at an early stage of the testing process can be utilized for several test optimization purposes, such as test case selection, prioritization, scheduling,and also parallel test execution. In this paper, we apply, evaluate, and compare the performance of two deep learning algorithmsto detect the similarities between manual integration test cases. The feasibility of the mentioned algorithms is later examined in a Telecom domain by analyzing the test specifications of five different products in the product development unit at Ericsson AB in Sweden. The empirical evaluation indicates that utilizing deep learning algorithms for finding the similarities between manual integration test cases can lead to outstanding results., TESTOMAT Project - The Next Level of Test Automation

Finding a balance between testing goals and testing resources can be considered as a most challenging issue, therefore test optimization plays a vital role in the area of software testing. Several parameters such as the objectives of the tests, test cases similarities and dependencies between test cases need to be considered, before attempting any optimization approach. However, analyzing corresponding testing artifacts (e.g. requirement specification, test cases) for capturing the mentioned parameters is a complicated task especially in a manual testing procedure, where the test cases are documented as a natural text written by a human. Thus, utilizing artificial intelligence techniques in the process of analyzing complex and sometimes ambiguous test data, is considered to be working in different industries. Test scheduling is one of the most popular and practical ways to optimize the testing process. Having a group of test cases which are required the same system setup, installation or testing the same functionality can lead to a more efficient testing process. In this paper, we propose, apply and evaluate a natural language processing-based approach that derives test cases' similarities directly from their test specification. The proposed approach utilizes the Levenshtein distance and converts each test case into a string. Test cases are then grouped into several clusters based on their similarities. Finally, a set of cluster-based parallel test scheduling strategies are proposed for execution. The feasibility of the proposed approach is studied by an empirical evaluation that has been performed on a Telecom use-case at Ericsson in Sweden and indicates promising results.

In this chapter we focus the use of knowledge representation and reasoning (KRR) methods as a guide to machine learning algorithms whereby relevant contextual knowledge can be leveraged upon. In this way, the learning methods improve performance by taking into account causal relationships behind errors. Performance improvement can be obtained by focusing the learning task on aspects that are particularly challenging (or prone to error), and then using added knowledge inferred by the reasoner as a means to provide further input to learning algorithms. Said differently, the KRR algorithms guide the learning algorithms, feeding it labels and data in order to iteratively reduce the errors calculated by a given cost function. This closed loop system comes with the added benefit that errors are also made more understandable to the human, as it is the task of the KRR system to contextualize the errors from the ML algorithm in accordance with its knowledge model. This represents a type of explainable AI that is focused on interpretability. This chapter will discuss the benefits of using KRR methods with ML methods in this way, and demonstrate an approach applied to satellite data for the purpose of improving classification and segmentation task.

Proper mapping and classification of Forest cover types are integral in understanding the processes governing the interaction mechanism of the surface with the atmosphere. In the presence of massive satellite and aerial measurements, a proper manual categorization has become a tedious job. In this study, we implement three different modest machine learning classifiers along with three statistical feature selectors to classify different cover types from cartographic variables. Our results showed that, among the chosen classifiers, the standard Random Forest Classifier together with Principal Components performs exceptionally well, not only in overall assessment but across all seven categories. Our results are found to be significantly better than existing studies involving more complex Deep Learning models.

Understanding why machine learning algorithms may fail is usually the task of the human expert that uses domain knowledge and contextual information to discover systematic shortcomings in either the data or the algorithm. In this paper, we propose a semantic referee, which is able to extract qualitative features of the errors emerging from deep machine learning frameworks and suggest corrections. The semantic referee relies on ontological reasoning about spatial knowledge in order to characterize errors in terms of their spatial relations with the environment. Using semantics, the reasoner interacts with the learning algorithm as a supervisor. In this paper, the proposed method of the interaction between a neural network classifier and a semantic referee shows how to improve the performance of semantic segmentation for satellite imagery data., Funding Agency:Swedish Knowledge Foundation under the research profile on Semantic Robots 20140033, Semantic Robot

Computed tomography (CT) is the method of choice for diagnosing ureteral stones - kidney stones that obstruct the ureter. The purpose of this study is to develop a computer aided detection (CAD) algorithm for identifying a ureteral stone in thin slice CT volumes. The challenge in CAD for urinary stones lies in the similarity in shape and intensity of stones with non-stone structures and how to efficiently deal with large high-resolution CT volumes. We address these challenges by using a Convolutional Neural Network (CNN) that works directly on the high resolution CT volumes. The method is evaluated on a large data base of 465 clinically acquired high-resolution CT volumes of the urinary tract with labeling of ureteral stones performed by a radiologist. The best model using 2.5D input data and anatomical information achieved a sensitivity of 100% and an average of 2.68 false-positives per patient on a test set of 88 scans., Funding Agencies:Nyckelfonden OLL-597511 Vinnova under the project "Interactive Deep Learning for 3D image analysis"

Machine learning algorithms, despite their increasing success in handling object recognition tasks, still seldom perform without error. Often the process of understanding why the algorithm has failed is the task of the human who, using domain knowledge and contextual information, can discover systematic shortcomings in either the data or the algorithm. This paper presents an approach where the process of reasoning about errors emerging from a machine learning framework is automated using symbolic techniques. By utilizing spatial and geometrical reasoning between objects in a scene, the system is able to describe misclassified regions in relation to its context. The system is demonstrated in the remote sensing domain where objects and entities are detected in satellite images., IJCAI Workshop on Learning and Reasoning: Principles & Applications to Everyday Spatial and Temporal Knowledge

Purpose: To develop a deep learning method for assisting radiologists in the discrimination between distal ureteral stones and pelvic phleboliths in thin slice CT images, and to evaluate whether this differentiation is possible using only local features. Methods and materials: A limited field-of-view image data bank was retrospectively created, consisting of 5x5x5 cm selections from 1 mm thick unenhanced CT images centered around 218 pelvis phleboliths and 267 distal ureteral stones in 336 patients. 50 stones and 50 phleboliths formed a validation cohort and the remainder a training cohort. Ground truth was established by a radiologist using the complete CT examination during inclusion.The limited field-of-view CT stacks were independently reviewed and classified as containing a distal ureteral stone or a phlebolith by seven radiologists. Each cropped stack consisted of 50 slices (5x5 cm field-of-view) and was displayed in a standard PACS reading environment. A convolutional neural network using three perpendicular images (2.5D-CNN) from the limited field-of-view CT stacks was trained for classification. Results: The 2.5D-CNN obtained 89% accuracy (95% confidence interval 81%-94%) for the classification in the unseen validation cohort while the accuracy of radiologists reviewing the same cohort was 86% (range 76%-91%). There was no statistically significant difference between 2.5D-CNN and radiologists. Conclusion: The 2.5D-CNN achieved radiologist level classification accuracy between distal ureteral stones and pelvic phleboliths when only using the local features. The mean accuracy of 86% for radiologists using limited field-of-view indicates that distant anatomical information that helps identifying the ureter’s course is needed.

In this paper we propose an ontology pattern that represents paths in a geo-representation model to be used in an aerial path planning processes. This pattern provides semantics related to constraints (i.e., ight forbidden zones) in a path planning problem in order to generate collision free paths. Our proposed approach has been applied on an ontology containing geo-regions extracted from satellite imagery data from a large urban city as an illustrative example., Semantic Robot

The availability of high-resolution remote sensing (HRRS) data has opened up the possibility for new interesting applications, such as per-pixel classification of individual objects in greater detail. This paper shows how a convolutional neural network (CNN) can be applied to multispectral orthoimagery and a digital surface model (DSM) of a small city for a full, fast and accurate per-pixel classification. The predicted low-level pixel classes are then used to improve the high-level segmentation. Various design choices of the CNN architecture are evaluated and analyzed. The investigated land area is fully manually labeled into five categories (vegetation, ground, roads, buildings and water), and the classification accuracy is compared to other per-pixel classification works on other land areas that have a similar choice of categories. The results of the full classification and segmentation on selected segments of the map show that CNNs are a viable tool for solving both the segmentation and object recognition task for remote sensing data.

There is an increasing interest in the machine learning community to automatically learn feature representations directly from the (unlabeled) data instead of using hand-designed features. The autoencoder is one method that can be used for this purpose. However, for data sets with a high degree of noise, a large amount of the representational capacity in the autoencoder is used to minimize the reconstruction error for these noisy inputs. This paper proposes a method that improves the feature learning process by focusing on the task relevant information in the data. This selective attention is achieved by weighting the reconstruction error and reducing the influence of noisy inputs during the learning process. The proposed model is trained on a number of publicly available image data sets and the test error rate is compared to a standard sparse autoencoder and other methods, such as the denoising autoencoder and contractive autoencoder.

This paper gives a review of the recent developments in deep learning and unsupervised feature learning for time-series problems. While these techniques have shown promise for modeling static data, such as computer vision, applying them to time-series data is gaining increasing attention. This paper overviews the particular challenges present in time-series data and provides a review of the works that have either applied time-series data to unsupervised feature learning algorithms or alternatively have contributed to modifications of feature learning algorithms to take into account the challenges present in time-series data.

This paper investigates a rapid and accurate detection system for spoilage in meat. We use unsupervised feature learning techniques (stacked restricted Boltzmann machines and auto-encoders) that consider only the transient response from undoped zinc oxide, manganese-doped zinc oxide, and fluorine-doped zinc oxide in order to classify three categories: the type of thin film that is used, the type of gas, and the approximate ppm-level of the gas. These models mainly offer the advantage that features are learned from data instead of being hand-designed. We compare our results to a feature-based approach using samples with various ppm level of ethanol and trimethylamine (TMA) that are good markers for meat spoilage. The result is that deep networks give a better and faster classification than the feature-based approach, and we thus conclude that the fine-tuning of our deep models are more efficient for this kind of multi-label classification task., Fuding agency: Department of Science & Technology, India

The main objective of an auto-encoder is to reconstruct the input signals via a feature representation of latent variables. The number of latent variables defines the representational capacity limit of the model. For data sets where some or all signals contain noise there is an unnecessary amount of capacity spent on trying to reconstruct these signals. One solution is to increase the number of hidden units to increase the capacity so that there will be enough capacity to capture the valuable information. Another solution is to pre-process the signals or perform a manual signal selection. In this paper, we propose a method that will dynamically change the objective function depending on the current performance of the model. This is done by weighting the objective function individually for each input unit in order to guide the feature leaning and decrease the influence that problematic signals have on the learning of features. We evaluate our method on various multidimensional time-series data sets and handwritten digit recognition data sets and compare our results with a standard sparse auto-encoder.

Most attempts at training computers for the difficult and time-consuming task of sleep stage classification involve a feature extraction step. Due to the complexity of multimodal sleep data, the size of the feature space can grow to the extent that it is also necessary to include a feature selection step. In this paper, we propose the use of an unsupervised feature learning architecture called deep belief nets (DBNs) and show how to apply it to sleep data in order to eliminate the use of handmade features. Using a postprocessing step of hidden Markov model (HMM) to accurately capture sleep stage switching, we compare our results to a feature-based approach. A study of anomaly detection with the application to home environment data collection is also presented. The results using raw data with a deep architecture, such as the DBN, were comparable to a feature-based approach when validated on clinical datasets.