Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.0-156
Églen Protas, José Douglas Bratti, J. O. Gaya, Paulo L. J. Drews-Jr, S. Botelho
In recent years, Convolutional Neural Networks (CNNs) have achieved state-of-the-art performance in many image restoration applications. The knowledge of how these models work, however, is still limited. While there have been many attempts at better understanding the inner working of CNNs, they have mostly been applied to classification networks. Because of this, most existing CNN visualization techniques may be inadequate to the study of image restoration architectures. In the paper, we present network inversion, a new method developed specifically to help in the understanding of image restoration Convolutional Neural Networks. We apply our method to underwater image restoration and dehazing CNNs, showing how it can help in the understanding and improvement of these models.
{"title":"Understading Image Restoration Convolutional Neural Networks with Network Inversion","authors":"Églen Protas, José Douglas Bratti, J. O. Gaya, Paulo L. J. Drews-Jr, S. Botelho","doi":"10.1109/ICMLA.2017.0-156","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.0-156","url":null,"abstract":"In recent years, Convolutional Neural Networks (CNNs) have achieved state-of-the-art performance in many image restoration applications. The knowledge of how these models work, however, is still limited. While there have been many attempts at better understanding the inner working of CNNs, they have mostly been applied to classification networks. Because of this, most existing CNN visualization techniques may be inadequate to the study of image restoration architectures. In the paper, we present network inversion, a new method developed specifically to help in the understanding of image restoration Convolutional Neural Networks. We apply our method to underwater image restoration and dehazing CNNs, showing how it can help in the understanding and improvement of these models.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"92 1","pages":"215-220"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77499551","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.0-110
Abdulaziz Almalaq, G. Edwards
The utility industry has invested widely in smart grid (SG) over the past decade. They considered it the future electrical grid while the information and electricity are delivered in two-way flow. SG has many Artificial Intelligence (AI) applications such as Artificial Neural Network (ANN), Machine Learning (ML) and Deep Learning (DL). Recently, DL has been a hot topic for AI applications in many fields such as time series load forecasting. This paper introduces the common algorithms of DL in the literature applied to load forecasting problems in the SG and power systems. The intention of this survey is to explore the different applications of DL that are used in the power systems and smart grid load forecasting. In addition, it compares the accuracy results RMSE and MAE for the reviewed applications and shows the use of convolutional neural network CNN with k-means algorithm had a great percentage of reduction in terms of RMSE.
{"title":"A Review of Deep Learning Methods Applied on Load Forecasting","authors":"Abdulaziz Almalaq, G. Edwards","doi":"10.1109/ICMLA.2017.0-110","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.0-110","url":null,"abstract":"The utility industry has invested widely in smart grid (SG) over the past decade. They considered it the future electrical grid while the information and electricity are delivered in two-way flow. SG has many Artificial Intelligence (AI) applications such as Artificial Neural Network (ANN), Machine Learning (ML) and Deep Learning (DL). Recently, DL has been a hot topic for AI applications in many fields such as time series load forecasting. This paper introduces the common algorithms of DL in the literature applied to load forecasting problems in the SG and power systems. The intention of this survey is to explore the different applications of DL that are used in the power systems and smart grid load forecasting. In addition, it compares the accuracy results RMSE and MAE for the reviewed applications and shows the use of convolutional neural network CNN with k-means algorithm had a great percentage of reduction in terms of RMSE.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"19 1","pages":"511-516"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84510595","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.0-174
Maggie B. Wigness, Steven Gutstein
State of the art visual classification continues to improve, particularly with the use of deep learning and millions of labeled images. However, the effort required to label training sets of this size has led to semi-supervised approaches that collect partially noisy labeled data with less effort. Label noise has been shown to degrade supervised learning, but these analyses focus on noise from erroneous label assignment of data instances. Group-based labeling reduces workload by assigning a single label to a group of images simultaneously, which introduces label noise with structure dependent on all training instances. This work investigates the impact of group-based label noise on classifier learning, and discusses how and why this differs from instance-based label noise. We also discuss label noise modeling designed to provide more robust classification given noisy training instances, and evaluate the generalization of these techniques to group-based noise.
{"title":"On the Impacts of Noise from Group-Based Label Collection for Visual Classification","authors":"Maggie B. Wigness, Steven Gutstein","doi":"10.1109/ICMLA.2017.0-174","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.0-174","url":null,"abstract":"State of the art visual classification continues to improve, particularly with the use of deep learning and millions of labeled images. However, the effort required to label training sets of this size has led to semi-supervised approaches that collect partially noisy labeled data with less effort. Label noise has been shown to degrade supervised learning, but these analyses focus on noise from erroneous label assignment of data instances. Group-based labeling reduces workload by assigning a single label to a group of images simultaneously, which introduces label noise with structure dependent on all training instances. This work investigates the impact of group-based label noise on classifier learning, and discusses how and why this differs from instance-based label noise. We also discuss label noise modeling designed to provide more robust classification given noisy training instances, and evaluate the generalization of these techniques to group-based noise.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"16 1","pages":"84-91"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76611206","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.0-141
S. Aminikhanghahi, Ramin Fallahzadeh, M. Sawyer, D. Cook, L. Holder
Smartphone prompts and notifications are popular because they provide users with timely and important information. However, they can also be an annoyance if they pop up at inopportune times and interrupt important tasks. In this paper, we introduce Thyme, an intelligent notification front end that uses activity recognition and machine learning to identify the best times to prompt smartphone users. We evaluate the performance of an activity-aware prompting approach based on 47 participants with fixed time and Thyme-based prompts. Our results show that responsiveness improves from 12.8% to 93.2% using this intelligent approach to the timing of smartphone-based prompts.
{"title":"Thyme: Improving Smartphone Prompt Timing Through Activity Awareness","authors":"S. Aminikhanghahi, Ramin Fallahzadeh, M. Sawyer, D. Cook, L. Holder","doi":"10.1109/ICMLA.2017.0-141","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.0-141","url":null,"abstract":"Smartphone prompts and notifications are popular because they provide users with timely and important information. However, they can also be an annoyance if they pop up at inopportune times and interrupt important tasks. In this paper, we introduce Thyme, an intelligent notification front end that uses activity recognition and machine learning to identify the best times to prompt smartphone users. We evaluate the performance of an activity-aware prompting approach based on 47 participants with fixed time and Thyme-based prompts. Our results show that responsiveness improves from 12.8% to 93.2% using this intelligent approach to the timing of smartphone-based prompts.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"87 1","pages":"315-322"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78262409","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.00014
B. O. Odelowo, David V. Anderson
In this paper, we propose a method for post-processing of deep neural network (DNN) enhanced speech. The method, which is simple and does not require additional training or expansion of the feature or target vectors, can be viewed as a mask-based approach in which a noisy speech signal is processed by a time-frequency (T-F) weighting derived from the noise-free spectral estimate of a DNN. A series of experiments and statistical analyses of results are carried out to compare the performance of the proposed approach to a baseline DNN enhancement system that features no post processing. Objective tests show that the proposed approach always improves both speech quality and intelligibility, and it outperforms a corresponding baseline system in both matched and mismatched noise conditions. Analysis of the enhanced speech shows that post processing reduces severe amplification distortions in the magnitude spectrum of the enhanced speech at the cost of a slight increase in severe attenuation distortions.
{"title":"A Mask-Based Post Processing Approach for Improving the Quality and Intelligibility of Deep Neural Network Enhanced Speech","authors":"B. O. Odelowo, David V. Anderson","doi":"10.1109/ICMLA.2017.00014","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.00014","url":null,"abstract":"In this paper, we propose a method for post-processing of deep neural network (DNN) enhanced speech. The method, which is simple and does not require additional training or expansion of the feature or target vectors, can be viewed as a mask-based approach in which a noisy speech signal is processed by a time-frequency (T-F) weighting derived from the noise-free spectral estimate of a DNN. A series of experiments and statistical analyses of results are carried out to compare the performance of the proposed approach to a baseline DNN enhancement system that features no post processing. Objective tests show that the proposed approach always improves both speech quality and intelligibility, and it outperforms a corresponding baseline system in both matched and mismatched noise conditions. Analysis of the enhanced speech shows that post processing reduces severe amplification distortions in the magnitude spectrum of the enhanced speech at the cost of a slight increase in severe attenuation distortions.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"35 1","pages":"1134-1138"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81362497","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.0-185
Wei Zhang, Rahul Sengupta, John Fodero, Xiaolin Li
Since WiFi has been pervasively available indoor, most smartphone indoor localization systems are based on WiFi fingerprinting although they only give coarse-grained location estimation. In this paper, we propose a novel deep learning-based indoor fingerprinting system (called DeepPositioning), combining Received Signal Strength Indicator (RSSI) of WiFi and pervasive magnetic field to obtain richer fingerprinting. DeepPositioning includes an offline learning phase and an online serving phase. In the offline learning phase, deep learning is utilized to automatically extract rich intrinsic features from a large number of multi-class fingerprints collected using mobile phones. Experimental results demonstrate that deep learning models with the intelligent fusion of pervasive WiFi and magnetic field data can effectively improve smartphone indoor localization compared to existing approaches based on WiFi only.
{"title":"DeepPositioning: Intelligent Fusion of Pervasive Magnetic Field and WiFi Fingerprinting for Smartphone Indoor Localization via Deep Learning","authors":"Wei Zhang, Rahul Sengupta, John Fodero, Xiaolin Li","doi":"10.1109/ICMLA.2017.0-185","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.0-185","url":null,"abstract":"Since WiFi has been pervasively available indoor, most smartphone indoor localization systems are based on WiFi fingerprinting although they only give coarse-grained location estimation. In this paper, we propose a novel deep learning-based indoor fingerprinting system (called DeepPositioning), combining Received Signal Strength Indicator (RSSI) of WiFi and pervasive magnetic field to obtain richer fingerprinting. DeepPositioning includes an offline learning phase and an online serving phase. In the offline learning phase, deep learning is utilized to automatically extract rich intrinsic features from a large number of multi-class fingerprints collected using mobile phones. Experimental results demonstrate that deep learning models with the intelligent fusion of pervasive WiFi and magnetic field data can effectively improve smartphone indoor localization compared to existing approaches based on WiFi only.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"134 1","pages":"7-13"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81652312","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.0-127
Mattyws F. Grawe, C. A. Martins, Andreia Gentil Bonfante
Patent classification is the task of assign a special code to a patent, where the assigned code is used to group patents with similar subject into a same category. This paper presents a patent categorization method based on word embedding and long short term memory network to classify patents down to the subgroup IPC level. The experimental results indicate that our classification method achieve 63% accuracy at the subgroup level.
{"title":"Automated Patent Classification Using Word Embedding","authors":"Mattyws F. Grawe, C. A. Martins, Andreia Gentil Bonfante","doi":"10.1109/ICMLA.2017.0-127","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.0-127","url":null,"abstract":"Patent classification is the task of assign a special code to a patent, where the assigned code is used to group patents with similar subject into a same category. This paper presents a patent categorization method based on word embedding and long short term memory network to classify patents down to the subgroup IPC level. The experimental results indicate that our classification method achieve 63% accuracy at the subgroup level.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"29 1","pages":"408-411"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87142957","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.0-111
Carmen Haseltine, E. Eman
Power Grid failures have the potential to drastically affect the population be it a localized outage or a large-scale blackout. Pre-event planning currently consists of preparation for all scenarios and some enthusiastic prognoses, leading to most resources spreading thin. Focus on a specific area of concern typically follows large scale power grid failures as post event analysis and does not include an overall analysis. In this study, a neural network is used to conduct “pre-event” analysis of a power grid to determine if it is susceptible to failure. This research study demonstrates that overall “pre-event” analysis can be beneficial with the use of a machine learning agent. The agent can also be used to determine areas that need the most attention. Future work with larger number of constraints and additional machine learning algorithms will be explored to further improve power grid analysis and performance.
{"title":"Prediction of Power Grid Failure Using Neural Network Learning","authors":"Carmen Haseltine, E. Eman","doi":"10.1109/ICMLA.2017.0-111","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.0-111","url":null,"abstract":"Power Grid failures have the potential to drastically affect the population be it a localized outage or a large-scale blackout. Pre-event planning currently consists of preparation for all scenarios and some enthusiastic prognoses, leading to most resources spreading thin. Focus on a specific area of concern typically follows large scale power grid failures as post event analysis and does not include an overall analysis. In this study, a neural network is used to conduct “pre-event” analysis of a power grid to determine if it is susceptible to failure. This research study demonstrates that overall “pre-event” analysis can be beneficial with the use of a machine learning agent. The agent can also be used to determine areas that need the most attention. Future work with larger number of constraints and additional machine learning algorithms will be explored to further improve power grid analysis and performance.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"22 9 1","pages":"505-510"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88253900","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.00-24
Sonu Agarwal, Sanjay Ghosh
Gesture based interaction and its recognition has been an area of active research with the growing popularity of wearables. We here propose an approach to detect fine-grained finger and palm motions using inertial sensors in a commercial smartwatch. A user specific SVM based classifier is developed for 7 microgestures with a classification accuracy of 94.4%. We extend this to a user adaptive model by including a few representative instances of a new user and achieve a classification accuracy of 91.7%. Further, we are able to differentiate between variations of a microgesture using three fundamental building blocks - distance, speed and orientation. A novel regression based approach is presented to predict the distance parameter. The idea is demonstrated on a swipe gesture with an error of 14%.
{"title":"Evaluation of Microgesture Recognition Using a Smartwatch","authors":"Sonu Agarwal, Sanjay Ghosh","doi":"10.1109/ICMLA.2017.00-24","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.00-24","url":null,"abstract":"Gesture based interaction and its recognition has been an area of active research with the growing popularity of wearables. We here propose an approach to detect fine-grained finger and palm motions using inertial sensors in a commercial smartwatch. A user specific SVM based classifier is developed for 7 microgestures with a classification accuracy of 94.4%. We extend this to a user adaptive model by including a few representative instances of a new user and achieve a classification accuracy of 91.7%. Further, we are able to differentiate between variations of a microgesture using three fundamental building blocks - distance, speed and orientation. A novel regression based approach is presented to predict the distance parameter. The idea is demonstrated on a swipe gesture with an error of 14%.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"2012 1","pages":"986-991"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82605410","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-12-01DOI: 10.1109/ICMLA.2017.00-55
M. Ziaeefard, R. Bergevin, Jean-François Lalonde
We propose a deep learning framework to analyse the uncertainty associated with dyadic human activities at a small temporal granularity. Such time-slice analysis is able to infer human behaviours from short-term observations. Instead of classifying time-slices into k classes of activities, we report to what degree of certainty each activity is occurring from definitely not occurring to definitely occurring. To this end, we extract CNN-based unary probabilities and pairwise relations between body joints. The unary term gives cues on the local appearance while the pairwise term captures the contextual relations between the parts. We extract the features from each frame in a timeslice and examine different temporal aggregation schemes to generate a descriptor for the whole time-slice. Evaluations on the TAP dataset which is well-suited for time-slice activity analysis demonstrate the effectiveness of our approach for the task of uncertainty analysis in activity prediction.
{"title":"Deep Uncertainty Interpretation in Dyadic Human Activity Prediction","authors":"M. Ziaeefard, R. Bergevin, Jean-François Lalonde","doi":"10.1109/ICMLA.2017.00-55","DOIUrl":"https://doi.org/10.1109/ICMLA.2017.00-55","url":null,"abstract":"We propose a deep learning framework to analyse the uncertainty associated with dyadic human activities at a small temporal granularity. Such time-slice analysis is able to infer human behaviours from short-term observations. Instead of classifying time-slices into k classes of activities, we report to what degree of certainty each activity is occurring from definitely not occurring to definitely occurring. To this end, we extract CNN-based unary probabilities and pairwise relations between body joints. The unary term gives cues on the local appearance while the pairwise term captures the contextual relations between the parts. We extract the features from each frame in a timeslice and examine different temporal aggregation schemes to generate a descriptor for the whole time-slice. Evaluations on the TAP dataset which is well-suited for time-slice activity analysis demonstrate the effectiveness of our approach for the task of uncertainty analysis in activity prediction.","PeriodicalId":6636,"journal":{"name":"2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"37 1","pages":"822-825"},"PeriodicalIF":0.0,"publicationDate":"2017-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82764983","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：