We argue that better use of "digital footprints" (data generated from students’ learning activities) could be used to improve the traditional lecture. We point out some potentially important data sources, a...
详细信息
This paper explores object detection in the small data regime, where only a limited number of annotated bounding boxes are available due to data rarity and annotation expense. This is a common challenge today with mac...
详细信息
ISBN:
(纸本)9781728148038
This paper explores object detection in the small data regime, where only a limited number of annotated bounding boxes are available due to data rarity and annotation expense. This is a common challenge today with machinelearning being applied to many new tasks where obtaining training data is more challenging, e.g. in medical images with rare diseases that doctors sometimes only see once in their life-time. In this work we explore this problem from a generative modeling perspective by learning to generate new images with associated bounding boxes, and using these for training an object detector. We show that simply training previously proposed generative models does not yield satisfactory performance due to them optimizing for image realism rather than object detection accuracy. To this end we develop a new model with a novel unrolling mechanism that jointly optimizes the generative model and a detector such that the generated images improve the performance of the detector. We show this method outperforms the state of the art on two challenging datasets, disease detection and small data pedestrian detection, improving the average precision on NIH Chest X-ray by a relative 20% and localization accuracy by a relative 50%.
From the newsfeeds, the products and services advertised to us, job screening, risk analysis, facial recognition and to the results we get through search engines, human-curated algorithms sitting behind the scenes, ar...
详细信息
Problem based learning model is applicable in applied Mathematics II course in order to obtain better students' problem-solving skill in mathematic. In this research, one can learn the differences of student perfo...
详细信息
Problem based learning model is applicable in applied Mathematics II course in order to obtain better students' problem-solving skill in mathematic. In this research, one can learn the differences of student performance to solve problem by problem-based learning compared to traditional learning. Besides, observation has been done to student activities during class session. data analysis shows significant of a = 0.05 gained t(count) = 2.16 and t(Table) = 1.72 in other hand teaaat > ttabie, one can conclude that problem-based learning increase student performance in solving problem compared to those in control group in applied Mathematics II course, Study Program of Industrial Engineering Polytechnic of South Aceh. The student activity was rate at 69.64% average.
In-hospital cardiac arrest (IHCA) diminish the survival rate of patients, despite most of the IHCA cases are preventable. More than 54% IHCA patient had abnormal clinical manifestation before they suffered a cardiac a...
详细信息
ISBN:
(纸本)9781538691380
In-hospital cardiac arrest (IHCA) diminish the survival rate of patients, despite most of the IHCA cases are preventable. More than 54% IHCA patient had abnormal clinical manifestation before they suffered a cardiac arrest. If appropriate steps were taken, patients' survival rate would be higher and medical expense would be decreased. This paper proposes a novel approach to detect IHCA before the event occurred. We construct two types of shifting windows (corresponding to two tasks) that allow machinelearning to be applied for our dataset which is severely imbalanced. The results show that our approach can effectively handle the imbalanced dataset for detecting cardiac arrest. As the selection of performance index, we used the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC). In our experiments, the best classifier is random forest for task 1, with AUROC of 0.88. LSTM is the best for task 2, with AUPRC of 0.71 for the second task.
In many situations, the choice of an adequate similarity measure or metric on the feature space dramatically determines the performance of machinelearning methods. Building automatically such measures is the specific...
详细信息
ISBN:
(纸本)9783030375997;9783030375980
In many situations, the choice of an adequate similarity measure or metric on the feature space dramatically determines the performance of machinelearning methods. Building automatically such measures is the specific purpose of metric/similarity learning. In [21], similarity learning is formulated as a pairwise bipartite ranking problem: ideally, the larger the probability that two observations in the feature space belong to the same class (or share the same label), the higher the similarity measure between them. From this perspective, the ROC curve is an appropriate performance criterion and it is the goal of this article to extend recursive tree-based ROC optimization techniques in order to propose efficient similarity learning algorithms. The validity of such iterative partitioning procedures in the pairwise setting is established by means of results pertaining to the theory of U-processes and from a practical angle, it is discussed at length how to implement them by means of splitting rules specifically tailored to the similarity learning task. Beyond these theoretical/methodological contributions, numerical experiments are displayed and provide strong empirical evidence of the performance of the algorithmic approaches we propose.
Physiological datasets such as Electroencephalography (EEG) data offer an insight into some of the less well understood aspects of human physiology. This paper investigates simple methods to develop models of high lev...
详细信息
ISBN:
(纸本)9783030329624;9783030329617
Physiological datasets such as Electroencephalography (EEG) data offer an insight into some of the less well understood aspects of human physiology. This paper investigates simple methods to develop models of high level behavior from low level electrode readings. These methods include using neuron activity based pruning and large time slices of the data. Both approaches lead to solutions whose performance and transparency are superior to existing methods.
Sign language is the non-verbal communication used by people with hearing and speaking impairments. The automatic recognition of sign languages is usually based on video analysis of the signer though this is difficult...
详细信息
ISBN:
(纸本)9781728146737
Sign language is the non-verbal communication used by people with hearing and speaking impairments. The automatic recognition of sign languages is usually based on video analysis of the signer though this is difficult when considering different light levels or the surrounding environment. The work in this paper uses electromyography (EMG) and focuses on letters of the Irish Sign Language (ISL) alphabet. EMG is the recording of the electrical activity produced to stimulate movement in the skeletal muscles. We capture muscle signals and inertial movement data using the Thalmic MYO armband and, in real time, recognise the ISL alphabet. Our implementation is based on signal processing, feature extraction and machinelearning. The only input required to translate the ISL gestures are EMG and movement data, thus our approach is usable in scenarios where using video for automatic recognition video is not possible.
The 2019international Symposium on Remote Sensing (ISRS-2019) took place in Taipei, Taiwan from 17 to 19 April 2019. ISRS is one of the distinguished conferences on the photogrammetry, remote sensing and spatial info...
The 2019international Symposium on Remote Sensing (ISRS-2019) took place in Taipei, Taiwan from 17 to 19 April 2019. ISRS is one of the distinguished conferences on the photogrammetry, remote sensing and spatial information sciences, especially in East Asia. More than 220 papers were presented in 37 technical sessions organized at the conference. This Special Issue publishes a limited number of featured peer-reviewed papers extended from their original contributions at ISRS-2019. The selected papers highlight a variety of topics pertaining to innovative concepts, algorithms and applications with geospatial sensors, systems, and data, in conjunction with emerging technologies such as artificial intelligence, machine leaning and advanced spatial analysis algorithms. The topics of the selected papers include the following: the on-orbit radiometric calibration of satellite optical sensors, environmental characteristics assessment with remote sensing, machinelearning-based photogrammetry and image analysis, and the integration of remote sensing and spatial analysis. The selected contributions also demonstrate and discuss various sophisticated applications in utilizing remote sensing, geospatial data, and technologies to address different environmental and societal issues. Readers should find the Special Issue enlightening and insightful for understanding state-of-the-art remote sensing and spatial information science research, development and applications.
In this paper, the malicious URLs detection is treated as a binary classification problem and performance of several well-known classifiers are tested with test data. The algorithms Random Forests and support Vector M...
详细信息
ISBN:
(纸本)9783030053666;9783030053659
In this paper, the malicious URLs detection is treated as a binary classification problem and performance of several well-known classifiers are tested with test data. The algorithms Random Forests and support Vector machine (SVM) are studied in particular which attain a high accuracy. These algorithms are used for training the dataset for classification of good and bad URLs. The dataset of URLs is divided into training and test data in 60:40, 70:30 and 80:20 ratios. Accuracy of Random Forests and SVMs is calculated for several iterations for each split ratio. According to the results, the split ratio 80:20 is observed as more accurate split and average accuracy of Random Forests is more than SVMs. SVM is observed to be more fluctuating than Random Forests in accuracy.
暂无评论