Are French Really That Different? Recognizing Europeans from Faces Using Data-Driven Learning

Size: px
Start display at page:

Download "Are French Really That Different? Recognizing Europeans from Faces Using Data-Driven Learning"

Transcription

1 Are French Really That Different? Recognizing Europeans from Faces Using Data-Driven Learning Viet-Duy Nguyen *, Minh Tran * and Jiebo Luo Department of Computer Science, University of Rochester, Rochester NY 14627, USA {vnguy14, mtran14}@u.rochester.edu, {jluo}@cs.rochester.edu Abstract Travel agents and retailers are curious about where their customers come from, which would help them increase their sales and optimize their marketing strategy. In this study, we present a system to predict where people come from in the European region only using their faces. The countries that have been chosen for the study are Russia, Italy, Germany, Spain, and France, based on diversity and representativeness. These countries have been well known for their economy, population, and political impact. First, we implement different neural network classifiers on the dataset of people s faces that we have collected from Twitter. Next, we investigate in more detail 11 different facial features that may help differentiate ethnic groups representative of those five countries. Our system achieves an accuracy of over 50%, more than twice as good as that of humans. Furthermore, we uncover and interpret using genetic anthropological evidences the various differences and similarities between people s faces across geographical distances among different contingents. I. INTRODUCTION From the following five images, can you tell the nationality of each person, given the choices are French, German, Italian, Russian, and Spanish? The answer, in order, is Russian, German, French, Italian, and Spanish. In fact, it is difficult to tell where people come from due to the relative distances between European countries. France, Germany, Italy, Russia and Spain are among the largest economies and populated countries in Europe 1,2, so these five countries have been chosen for the study. In the first stage of the study, we implement two different state-of-the-art neural network architectures, fine-tuned VGG- Net 16 [1] and fine-tuned ResNet50 [2]. On our collected Twitter dataset, with ResNet50, we achieve a remarkable accuracy of 53.2%, with the chances being 20%. Moreover, a user experiment has been conducted on 72+ individuals 3 and the human performance recorded so far is 26.96%. In the second stage of the study, we intend to explore informative facial features that help distinguish among European people. We use the CelebA Dataset [3] to train our model 11 times for 11 chosen facial attributes. We then use the trained models to classify images from our collected Twitter dataset. We have obtained interesting insights that can be inferred from our classification results. For example, Italians wear eyeglasses on their social media images more frequently than others, * Both authors contributed equally to this work The experiment is available at Fig. 1. It is challenging to tell the differences between German, Italian, Spanish, Russian, and French. while Russians rarely wear eyeglasses. Moreover, the analysis on cross-country gender differentials is also reported. Finally, we use the trained models in the first stage to classify people s faces from around the world. We choose as inputs to our classification: Turkey, Saudi Arabia and Japan for Asia; Egypt for Africa; Cuba and Peru for America. Literature in History and Genetic Anthropology has been used to provide explanation for the results we obtain; for example, Turkish people resemble Italians the most, and Japanese look like Russians the most among the five European countries. In summary, we make several contributions in this study. First, using machine learning techniques, we build a system for recognizing Europeans from faces, which is about twice as good as humans. Second, we present a novel way to explain ethnic groups that share similar Y-DNA haplogroup distribution and related history solely by the analysis on faces. II. RELATED WORKS The study is related to the topics of deep learning for face and facial attribute classification, as well as genetic anthropology. In the field of face classification by neural networks, there are successes in terms of accuracy [4] thanks to large datasets [5], [6], [7]. The problem of race classification based on face images has been successfully tackled with an accuracy up to 98% as summarized by Fu et al.[8]. A few architectures are proposed [1], [2], [9], [10], [11]. Deep learning has been applied to not only face classification but also facial attribute classification. [12] is one of the earliest works in the field. [3] has achieved a great performance by training two networks. The first is for face localization and the second is for attribute classification. Following that work, [13] works well on a similar dataset with a different approach: using OpenCV to locate faces and training neural networks for every attribute. Our work is inspired by [13] for both face and facial attribute classification. We use VGG-Net16 and ResNet50 for face classification. [13] investigates data from three East

2 Asian countries (Japan, China, and Korea) with an accuracy of 75.03% for face classification. Our work has expanded the number of countries from 3 Asian countries to 5 European countries with the expectation of a lower accuracy due to the increased complexity. Instead of the neural network used in [13] for facial attribute classification, we train ResNet50 for each facial attribute. In the field of genetic anthropology, there is a study on the distribution of Y-DNA haplogroups in the populations of Europe [14]. The study suggests that unlike the other four countries, Russia tends to have most R1a and N in its population (46% and 23%, respectively). Although France, Germany, Spain and Italy have R1b as the haplogroup with the highest frequency, the second highest frequency haplogroup in these countries is different. For Spain, E1b1b is the second highest frequency haplogroup, while it is J2 for Italy. Both France and Germany have I1 as their second highest frequency haplogroup, but the frequency of I1 in Germany is about twice that of France (16% vs. 8.5%). This is clear evidence that there must be some measurable differences in facial appearance between the people from these five countries, which makes it possible to predict the nationalities of these people with a reasonable accuracy. Besides Europe, there exists much research on the distribution of Y-DNA haplogroups of countries all over the world. [15], [16] and [17] report the percentages of Y-DNA haplogroups in Japan, Saudi Arabia and Turkey, respectively. By studying the changes among different haplogroups, it is possible to project a map of migration flows of people from different countries around the world to Europe, or vice versa. III. DATASET This study uses two main datasets: Twitter images and CelebA dataset. The Twitter dataset contains labeled face images of 11 countries (France, Germany, Spain, Italy, Russia, Japan, Saudi Arabia, Turkey, Cuba, Peru and Egypt). The first five countries images are used to train the neural networks, while the others are used as the input of classification. The CelebA dataset is used to train the classification models for 11 facial features. A. Twitter Images The Twitter images are collected in a similar way for both Part 1 and Part 3 of the study. Part 1 of the study uses images of French, German, Spanish, Italian and Russian people. Part 3 of the study uses images of Saudi Arabian, Turkish, Egyptian, Japanese, Cuban and Peruvian people. Though the data is not ultimately clean, we tried our best to reduce cases that somebody of mixed descent by carefully choosing local celebrities or TV shows to retrieve data. First, we pick an influential Twitter account for each country and collect the followers of these accounts: French centrist politician Francois Bayrou for France, German reality talent show Deutschland sucht den Superstar for Germany, Spanish politician Mariano Rajoy for Spain, Italian Prime Minister Paolo Gentiloni for Italy, Russian Ice Hockey Federation Fig. 2. Resnet 50 Training loss. The graph indicates that the convergence is achieved. Twitter page for Russia, Cuban blogger Yoani Mara Snchez Cordero for Cuba, the Official Twitter Account of the Presidency of the Republic of Turkey for Turkey, President of Egypt Abdel Fattah el-sisi for Egypt, President of Peru Pedro Pablo Kuczynski for Peru and Japanese actor Tsuyoshi Muro for Japan. Since the initial dataset at this stage contains much noise, we restrict the list of followers based on the language they use on Twitter. We detect the language from the followers names and profile descriptions by using Google Language APIs [18] and from their Twitter language settings. Next, only images containing exactly one face are kept [19]. These images dimensions are resized to 224x224 pixels to fit the standard input for the neural network classification models [9]. Finally, we remove all images that have a size lower than 25kB to ensure high-quality images in our dataset. There are 30,806 French images, 34,265 German images, 31,929 Spanish images, 28,033 Italian images, and 32,365 Russian images; plus 15,049 Japanese images, 22,260 Saudi Arabian images, 44,480 Turkish images, 29,837 Cuban images, 41,945 Peruvian images and 42,857 Egyptian images. Our Twitter data collection method is not reliable to collect people s faces from English-speaking countries. The reason is that English is a widely used language in the British Commonwealth, which would lead to a problem of discerning where an English speaker is from. Therefore, it is not effective to apply the method above to collect the people s faces from countries such as the United States, Britain, or Australia. That is why British is excluded from this study (but can be added later with a better data collection method). B. CelebA Dataset For the second stage of the study, we use the CelebA Dataset [3]. The CelebA Dataset is a large-scale face attribute dataset with 10,177 identities and 202,599 images. Each image contains 40 annotated-binary face attributes, 11 out of which are chosen for this study. They are Eyeglasses, No beard, Chubby, Smile, Bangs, Rosy cheek, High cheekbones, Narrow

3 Fig. 3. On the left is the confusion matrix of ResNet50. On the right, the first two columns of the table are the result of facial features classification, which shows that our model gives a better result than LNet+ANet [3] on most of the 11 attributes. The second two columns are the probabilities of individual features for both genders. eyes, Pointy nose, Bushy, and Heavy makeup. The resulting models will be used to classify the facial attributes of people from the five countries. IV. METHOD In the first two stages of the study, we employ the cross validation method and divide our dataset into the ratio of 8 : 1 : 1 corresponding to training set : validation set : test set. A. Face Classification 1) Analysis on European Faces: Fine-tuned models often give a significantly higher accuracy than the original models [20]. With that notion, we use pre-trained VGG-Net 16 and ResNet50 on ImageNet [21] to train on our dataset. We reach an accuracy of 28.1% for fine-tuned VGG-Net 16 and a higher accuracy of 53.2% for fine-tuned ResNet50 after 120,000 iterations. Moreover, all of the models have been trained on the Caffe Framework and converge. 2) Analysis on Countries outside Europe: We apply our trained ResNet50 model not only in the prediction task but also in analyzing people s faces from places outside Europe. Using the trained model, we obtain the probabilities of the five classes (Spanish, Russian, German, Italian, and French) for every person s face picture from the six countries in Asia, Africa, and America. For each country outside Europe, we calculate the percentage of pictures that have been classified into a specific country among our five initial European countries (Figure 6). B. Facial Attribute Classification Given that ResNet50 gives us a high accuracy in classifying people s faces, we train it on CelebA Dataset 11 times for 11 different facial attributes. The second column of the table in Figure 3 is the result of our training on the facial attributes. Next, we implement facial attribute classification on our collected Twitter dataset using the model. However, we realize that there is an imbalance between the proportion of males and females in this study (80% vs. 20%). Therefore, we decide to perform a cross-country gender differential analysis. Some features have very low probabilities to appear on both the male and female (the forth and fifth columns of the table in Figure 3), which might make it difficult to achieve a good generalization of the whole population for the five countries. Therefore, we set the threshold to be 0.05, and the facial attributes that satisfy this criteria are Chubby, No beard, Eyeglasses and Smile. V. FACIAL ATTRIBUTES From the results of the facial attribute classification, we continue to extract further insights about the differences between males and females on their Twitter profiles across the five countries on each of the three facial attributes (Chubby, Eyeglasses, and Smile). We try our best to demystify the findings from the study, though not all features are easy to explain and find literature support. A. Chubby First, we observe that, for both sexes, France, Spain, and Italy are the top three countries with chubby people, followed by Germany and Russia (Figure 4). The order fits the Prevalence of Overweight for Adult by WHO in 2016 [22]. Furthermore, we can see that the percentage of chubby males is much higher than that of females in all five countries (Figure 4, Figure 3). This is an evidence that women appear or portray themselves to look less chubby on their Twitter profiles. B. Eyeglasses First, the distribution of the Eyeglasses attribute for both sexes in Figure 4 can be explained by the cultures of the five countries. France, Italy and Spain are well-known countries for

4 Fig. 4. Cross-country gender differentials on Smile, Eyeglasses, and Chubby. Red represents male, while Blue represents female. Fig. 5. Map of possible migration flow of populations around the world based on inferences of differences between Y-DNA distributions of countries. The map is designed by Bert Chen and posted at their fashion industry, as reported in [23] as the three worldleading fashionable countries. Therefore, it is reasonable that citizens of these three countries wears more eyeglasses than Russian or German (countries that are not in top 10). Second, we have the same result as [13] that males tend to wear eyeglasses more than females for all five countries. The reason for this finding was suggested to be the difference between work-pressure of men and women. C. Smile The Spanish tend to smile the most, followed by Italian, French and German people (around the same level). Russians smile the least in their Twitter profile images. This finding matches the Happy Planet Index by New Economics Foundation (NEF) in 2016 [24]. Unlike previous attributes, the percentages of smiling females are always higher than those of males. Although there

5 is no concrete interpretation for this, it is likely that females prefer to look friendly while males may want to look cool on their Twitter profiles. VI. INTERCONTINENTAL FACIAL SIMILARITIES In the third stage, we use the ResNet50 model that we have trained in the first stage of the study. We have chosen different countries from Asia, America and Africa: Japan, Turkey, Saudi Arabia, Peru, Cuba, and Egypt. The data of the six countries from Asia, Africa, and America is used as the input to our model. Feeding the Twitter images that are prepared for Part 3 into the classification model of the five European countries, we obtain the following interesting results. A. Spanish look-alike From Figure 6, it is clear that Cuban and Peruvian people look Spanish, with more than 40% chance a Peruvian or Cuban person would be classified as a Spanish person. This could be explained due to the colonization by Spain in these two countries. Cuba was a Spanish colony for over four centuries ( ), beginning with the arrival of Spanish explorer Christopher Columbus and ending with the withdrawal of Spain from the island due to the Spanish-American War. Peru was a Spanish colony for four decades ( ). Therefore, both Cuba and Peru might share significant genetic similarities with Spain, which could be the reason why people from these two countries look Spanish. There are few Peruvians who have been classified as the other people, which can be explained by the various migrations into South America in the 18th-20th century. B. Italian look-alike There are two ways to explain why Saudi Arabian and Turkish people look similar to Italian people (Figure 6): Y- DNA Haplogroup Distributions and Historical Background. First, both Saudi Arabia and Turkey used to be colonized by the Roman Empire. In 106 CE, Roman Emperor Trajan took over the Arab Kingdom of the Nabataeans as the province of Arabia, which represented the Roman Empires easternmost frontier [25]. The Roman Empire also occupied Asia Minor, which makes up of the majority of Turkey, after Attalos III of Pergamon bequeathed his kingdom to Rome in 133 BCE [26]. These facts imply that there was a flow of people from Italia to both Turkey and Saudi Arabia. Moreover, the similarities between Italian and Turkish & Saudi Arabian people can be explained using the distribution of haplogroups in these countries. In Italy, the dominant Y-DNA groups are R1b, J2 and E1b1b, corresponding to 39%, 15.5% and 13.5% respectively [14]. In Saudi Arabia, the dominant haplogroups are E1b1a, E1b1b and J (including J1 and J2), which cover 7.6%, 7.6% and 58% of the population respectively [16]. In Turkey, the dominant haplogroups are R1b, E1b1b and J, corresponding to 16.1%, 11.3% and 33.3% of the population, respectively [17]. Because of the resemblance in the distribution of haplogroups between the three countries, people from these countries should look similar to each other. The world map in Figure 5 shows possible migration flows in the world based on the changes among different haplogroups in the world, which includes the flow from Italy to Saudi Arabia and Turkey (the blue dashed-line). C. Russian look-alike According to Figure 6, Japanese have the highest percentage of pictures classified as Russian. This matches the projected migration flow by Y-DNA haplogroup (Figure 5). The group of people in Northern Japan brought the C and D halopgroups to Southern and Middle Russia via the sea route. D. French look-alike The high percentage of Egyptian people being classified as French (31.07%) can be explained by the occupation of Egypt by France from In this period, Napoleon Bonaparte, a French General, took control of Egypt [27]. This is evidence that there were groups of people from those two countries migrating between each other. VII. CONCLUSION In summary, we have achieved an accuracy of 53.2% for classifying face images of French, German, Italian, Russian and Spanish people. A approach to hopefully further improving the accuracy is to use the fine-tuned DenseNet [28], which is a more up-to-date neural network architecture. Based on our exploration of the 11 different facial attributes, we have found the following interesting results about the people from the five countries: 1) For both sexes, France, Spain, and Italy are the top three countries that have chubby people, followed by Germany and Russia. 2) People from Italy, France and Spain tend to wear more eyeglasses in their Twitter profiles than people from Russia and Germany. 3) The Spanish tend to smile the most, followed by Italian, French and German people. The Russians smile the least in their Twitter profile images. 4) Men in all the five countries smile less than women in their Twitter profile images. 5) Men in all the five countries tend to wear eyeglasses more than women. Finally, we also study other countries outside the European region to see if they might share facial similarities using machine learning techniques. There are 6 countries outside Europe (Japan, Saudi Arabia, Turkey, Cuba, Peru, Egypt) which have been collected from Twitter for this stage of study. These images are processed using our trained ResNet50 model in the first stage. We discover interesting insight that can be explained by the historical background and Y-DNA halopgroups of the six countries compared to the five European countries (Germany, France, Italy, Russia, and Spain). ACKNOWLEDGMENTS We thank the support of New York State through the Goergen Institute for Data Science, and our corporate sponsor Cheetah Mobile.

6 Fig. 6. Pie Chart of the percentages of classes (French, German, Russian, Italian, and Spanish) present in each country outside Europe. REFERENCES [1] K. Simonyan and A. Zisserman, Very deep convolutional networks for large-scale image recognition. International Conference on Learning Representations, [2] K. He, X. Zhang, S. Ren, and J. Sun, Deep residual learning for image recognition. International Conference on Computer Vision and Pattern Recognition, [3] Z. Liu, P. Luo, X. Wang, and X. Tang, Deep learning face attributes in the wild. International Conference on Computer Vision, [4] Y. Wang, Y. Li, and J. Luo., Deciphering the 2016 u.s. presidential campaign in the twitter sphere: A comparison of the trumpists and clintonists. Tenth International AAAI Conference on Web and Social Media, [5] G. B. Huang, M. Ramesh, T. Berg, and E. Learned-Miller, Labeled faces in the wild: A database for studying face recognition in unconstrained environments, Technical report, University of Massachusetts, [6] P. J. Phillips, H. Wechslerb, J. Huangb, and P. J. Raussa, The feret database and evaluation procedure for face-recognition algorithms. Image and Vision Computing, pp , [7] Y. Guo, L. Zhang, Y. Hu, X. He, and J. Gao, Ms-celeb-1m: A dataset and benchmark for large-scale face recognition. European Conference on Computer Vision, [8] S. Fu, H. He, and Z.-G. Hou, Learning race from face: A survey, IEEE transactions on pattern analysis and machine intelligence, vol. 36, no. 12, pp , [9] A. Krizhevsky, I. Sutskever, and G. E. Hinton., Imagenet classification with deep convolutional neural networks. In Advances in Neural Information Processing Systems, [10] A. Romero, N. Ballas, S. E. Kahou, A. Chassang, C. Gatta, and Y. Bengio, Imagenet classification with deep convolutional neural networks. In International Conference on Learning Representations, [11] J. G. Zilly, R. K. Srivastava, J. Koutnik, and J. Schmidhuber, Recurrent highway networks. arxiv: , [12] N. Zhang, M. Paluri, M. Ranzato, T. Darrell, and L. Bourdev., Panda: Pose aligned networks for deep attribute modeling. International Conference on Computer Vision and Pattern Recognition, [13] Y. Wang, H. Liao, Y. Feng, and X. Xu, Do they all look the same? deciphering chinese, japanese and koreans by fine-grained deep learning, arxiv: , [14] Eupedia.com, Distribution of european y-chromosome dna (y-dna) haplogroups by country in percentage, [15] M. F. Hammer, T. M. Karafet, H. Park, K. Omoto, S. Harihara, M. Stoneking, and S. Horai, Dual origins of the japanese: common ground for hunter-gatherer and farmer y chromosomes, Journal of Human Genetics Volume 51, [16] K. K, H. Ali, G. A. M, L. J. M, C. V. M, and U. P. A, Saudi arabian y- chromosome diversity and its relationship with nearby regions, National Center for Biotechnology Information, [17] C. Cengiz, K. Roy, K. Toomas, K. Ersi, A. Sevil, C. G. L., L. A. S., and R. C. C., Excavating y-chromosome haplotype strata in anatolia, 114: [18] N. Shuyo, Language detection library for java [19] A. Geitgey, Modern Face Recognition with Deep Learning. Medium Corporation, [20] Y. LeCun, Y. Bengio, and G. Hinton, Deep learning, Nature, [21] J. Deng, W. Dong, R. Socher, K. Li, L.-J.and Li, and Fei-Fei, Imagenet: A large-scale hierarchical image database. Proceedings of the International Conference on Computer Vision and Pattern Recognition, [22] Overweight and obesity, World Health Organization, [23] R. Dicker, 10 most fashionable countries, US News, [24] K. Jeffrey, H. Wheatley, and S. Abdallah, The happy planet index, New Economics Foundation, [25] J. Dombrowski, Arabia, roman province, The Encyclopedia of Ancient History, [26] S. Mitchell, Asia, roman province of, The Encyclopedia of Ancient History, [27] C. Cherfils, Bonaparte et l Islam d apres les documents francais & arabes. Pedone, [28] G. Huang, Z. Liu, K. Q. Weinberger, and L. V. D. Maaten, Densely connected convolutional networks, arxiv: , 2016.

arxiv: v2 [cs.cv] 23 Oct 2016

arxiv: v2 [cs.cv] 23 Oct 2016 Do They All Look the Same? Deciphering Chinese, Japanese and Koreans by Fine-Grained Deep Learning Yu Wang, Haofu Liao, Yang Feng, Xiangyang Xu, Jiebo Luo* arxiv:1610.01854v2 [cs.cv] 23 Oct 2016 Department

More information

Multi-attention Guided Activation Propagation in CNNs

Multi-attention Guided Activation Propagation in CNNs Multi-attention Guided Activation Propagation in CNNs Xiangteng He and Yuxin Peng (B) Institute of Computer Science and Technology, Peking University, Beijing, China pengyuxin@pku.edu.cn Abstract. CNNs

More information

arxiv: v1 [stat.ml] 23 Jan 2017

arxiv: v1 [stat.ml] 23 Jan 2017 Learning what to look in chest X-rays with a recurrent visual attention model arxiv:1701.06452v1 [stat.ml] 23 Jan 2017 Petros-Pavlos Ypsilantis Department of Biomedical Engineering King s College London

More information

CSE Introduction to High-Perfomance Deep Learning ImageNet & VGG. Jihyung Kil

CSE Introduction to High-Perfomance Deep Learning ImageNet & VGG. Jihyung Kil CSE 5194.01 - Introduction to High-Perfomance Deep Learning ImageNet & VGG Jihyung Kil ImageNet Classification with Deep Convolutional Neural Networks Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton,

More information

When Saliency Meets Sentiment: Understanding How Image Content Invokes Emotion and Sentiment

When Saliency Meets Sentiment: Understanding How Image Content Invokes Emotion and Sentiment When Saliency Meets Sentiment: Understanding How Image Content Invokes Emotion and Sentiment Honglin Zheng1, Tianlang Chen2, Jiebo Luo3 Department of Computer Science University of Rochester, Rochester,

More information

Aim #1a: How do we analyze and interpret different types of data?

Aim #1a: How do we analyze and interpret different types of data? Aim #1a: How do we analyze and interpret different types of data? Data (information) may appear in various forms such as a map, graph, chart, table, timelines, political cartoons, visuals, and reading

More information

2016 HSC Modern History Marking Guidelines

2016 HSC Modern History Marking Guidelines 2016 HSC Modern History Marking Guidelines Section I World War I 1914 1919 Part A Multiple-choice Answer Key Question Answer 1 D 2 A 3 C 4 B 5 A 1 Question 6 Provides characteristics and features of ONE

More information

Comparison of Two Approaches for Direct Food Calorie Estimation

Comparison of Two Approaches for Direct Food Calorie Estimation Comparison of Two Approaches for Direct Food Calorie Estimation Takumi Ege and Keiji Yanai Department of Informatics, The University of Electro-Communications, Tokyo 1-5-1 Chofugaoka, Chofu-shi, Tokyo

More information

arxiv: v2 [cs.cv] 19 Dec 2017

arxiv: v2 [cs.cv] 19 Dec 2017 An Ensemble of Deep Convolutional Neural Networks for Alzheimer s Disease Detection and Classification arxiv:1712.01675v2 [cs.cv] 19 Dec 2017 Jyoti Islam Department of Computer Science Georgia State University

More information

An increasing demand for livestock directly translates into increasing demand for animal feed and animal

An increasing demand for livestock directly translates into increasing demand for animal feed and animal An increasing demand for livestock directly translates into increasing demand for animal feed and animal feed additives such as vitamins. Synthetic vitamin E is the key ingredient used in global animal

More information

The Test-Curriculum Matching Analysis: Science

The Test-Curriculum Matching Analysis: Science Appendix C The Test-Curriculum Matching Analysis: Science TIMSS went to great lengths to ensure that comparisons of student achievement across countries would be as fair and equitable as possible. The

More information

Object Detectors Emerge in Deep Scene CNNs

Object Detectors Emerge in Deep Scene CNNs Object Detectors Emerge in Deep Scene CNNs Bolei Zhou, Aditya Khosla, Agata Lapedriza, Aude Oliva, Antonio Torralba Presented By: Collin McCarthy Goal: Understand how objects are represented in CNNs Are

More information

Skin cancer reorganization and classification with deep neural network

Skin cancer reorganization and classification with deep neural network Skin cancer reorganization and classification with deep neural network Hao Chang 1 1. Department of Genetics, Yale University School of Medicine 2. Email: changhao86@gmail.com Abstract As one kind of skin

More information

Pandemic Preparedness

Pandemic Preparedness Pandemic Preparedness Disclosure Dr. Michael Braida is a full-time International SOS / MedAire employee Regional Medical Director Northern Seas based in Aberdeen MedAire Medical Director for Europe Medical

More information

arxiv: v1 [cs.cv] 17 Aug 2017

arxiv: v1 [cs.cv] 17 Aug 2017 Deep Learning for Medical Image Analysis Mina Rezaei, Haojin Yang, Christoph Meinel Hasso Plattner Institute, Prof.Dr.Helmert-Strae 2-3, 14482 Potsdam, Germany {mina.rezaei,haojin.yang,christoph.meinel}@hpi.de

More information

Appendix F: Test-Curriculum Matching Analysis

Appendix F: Test-Curriculum Matching Analysis Appendix F: Test-Curriculum Matching Analysis TIMSS went to great lengths to ensure that comparisons of student achievement across countries would be as fair and equitable as possible. The TIMSS 2015 Assessment

More information

Design of Palm Acupuncture Points Indicator

Design of Palm Acupuncture Points Indicator Design of Palm Acupuncture Points Indicator Wen-Yuan Chen, Shih-Yen Huang and Jian-Shie Lin Abstract The acupuncture points are given acupuncture or acupressure so to stimulate the meridians on each corresponding

More information

EPIDEMIOLOGY. Accurate, in-depth information for understanding and assessing targeted markets

EPIDEMIOLOGY. Accurate, in-depth information for understanding and assessing targeted markets EPIDEMIOLOGY Accurate, in-depth information for understanding and assessing targeted markets WHY ACCURATE MARKET FORECASTS Start with accurate understanding Epidemiology is the study of disease patterns

More information

Image Captioning using Reinforcement Learning. Presentation by: Samarth Gupta

Image Captioning using Reinforcement Learning. Presentation by: Samarth Gupta Image Captioning using Reinforcement Learning Presentation by: Samarth Gupta 1 Introduction Summary Supervised Models Image captioning as RL problem Actor Critic Architecture Policy Gradient architecture

More information

Visual semantics: image elements. Symbols Objects People Poses

Visual semantics: image elements. Symbols Objects People Poses Visible Partisanship Polmeth XXXIII, Rice University, July 22, 2016 Convolutional Neural Networks for the Analysis of Political Images L. Jason Anastasopoulos ljanastas@uga.edu (University of Georgia,

More information

ICGC launches new project and releases more genomic data on cancer

ICGC launches new project and releases more genomic data on cancer ICGC launches new project and releases more genomic data on cancer Toronto March 15, 2012. The International Cancer Genome Consortium (ICGC) today announced a new project from South Korea to identify the

More information

OVERVIEW. Denver International Airport 1

OVERVIEW. Denver International Airport 1 OVERVIEW Introduction Denver International Airport ( DEN or Airport ) is a large hub airport serving 58.3 million passengers each year through 572,520 (2016) operations (landings and takeoffs). It is the

More information

World Connections Committee (WCC) Report

World Connections Committee (WCC) Report World Connections Committee (WCC) Report 06 Co-Dependents Anonymous Service Conference Countries Where CoDA Exists This report reflects the World Connections Committee (WCC) support of the growth and development

More information

ArtTalk: Labeling Images with Thematic and Emotional Content

ArtTalk: Labeling Images with Thematic and Emotional Content ArtTalk: Labeling Images with Thematic and Emotional Content Yifan Wang Stanford University, VMware 3401 Hillview Ave, Palo Alto, CA, 94304 yifanwang@vmware.com Mana Lewis Chez Mana Co. mana@chezmana.com

More information

arxiv: v1 [cs.si] 27 Oct 2016

arxiv: v1 [cs.si] 27 Oct 2016 The Effect of Pets on Happiness: A Data-Driven Approach via Large-Scale Social Media arxiv:1610.09002v1 [cs.si] 27 Oct 2016 Yuchen Wu Electrical and Computer Engineering University of Rochester Rochester,

More information

ESMO ACROSS ONCOLOGY. WORLDWIDE. Josep Tabernero. ESMO President. ESMO SUMMIT MIDDLE EAST Friday, 6 th April 2018

ESMO ACROSS ONCOLOGY. WORLDWIDE. Josep Tabernero. ESMO President. ESMO SUMMIT MIDDLE EAST Friday, 6 th April 2018 ESMO ACROSS ONCOLOGY. WORLDWIDE Josep Tabernero ESMO President ESMO SUMMIT MIDDLE EAST Friday, 6 th April 2018 ESMO IN A NUTSHELL Europe s leading Medical Oncology Society ESMO is the leading European

More information

Cotton Trade 02/03 03/04 04/05 05/06 06/07 07/08 08/09 09/10 10/11 11/12 12/13

Cotton Trade 02/03 03/04 04/05 05/06 06/07 07/08 08/09 09/10 10/11 11/12 12/13 Cotton Trade Net Exporters Africa Argentina Australia India Other Former Soviet Union Other Latin America Other Middle East Pakistan Turkey United States Uzbekistan Total Net Exports 5.01 5.62 5.62 5.66

More information

Autumn-Winter Special Promotion

Autumn-Winter Special Promotion 5% Disco unt O ffer ROUTLEDGE REFERENCE Autumn-Winter Special Promotion Now through the end of the year, Routledge is offering a 5% discount on 50 carefully selected titles across the Humanities and Social

More information

Automated diagnosis of pneumothorax using an ensemble of convolutional neural networks with multi-sized chest radiography images

Automated diagnosis of pneumothorax using an ensemble of convolutional neural networks with multi-sized chest radiography images Automated diagnosis of pneumothorax using an ensemble of convolutional neural networks with multi-sized chest radiography images Tae Joon Jun, Dohyeun Kim, and Daeyoung Kim School of Computing, KAIST,

More information

arxiv: v1 [cs.lg] 4 Feb 2019

arxiv: v1 [cs.lg] 4 Feb 2019 Machine Learning for Seizure Type Classification: Setting the benchmark Subhrajit Roy [000 0002 6072 5500], Umar Asif [0000 0001 5209 7084], Jianbin Tang [0000 0001 5440 0796], and Stefan Harrer [0000

More information

Global EHS Resource Center

Global EHS Resource Center Global EHS Resource Center Understand environmental and workplace safety requirements that affect your global operations. 800.372.1033 bna.com/gelw Global EHS Resource Center This comprehensive research

More information

A convolutional neural network to classify American Sign Language fingerspelling from depth and colour images

A convolutional neural network to classify American Sign Language fingerspelling from depth and colour images A convolutional neural network to classify American Sign Language fingerspelling from depth and colour images Ameen, SA and Vadera, S http://dx.doi.org/10.1111/exsy.12197 Title Authors Type URL A convolutional

More information

A HMM-based Pre-training Approach for Sequential Data

A HMM-based Pre-training Approach for Sequential Data A HMM-based Pre-training Approach for Sequential Data Luca Pasa 1, Alberto Testolin 2, Alessandro Sperduti 1 1- Department of Mathematics 2- Department of Developmental Psychology and Socialisation University

More information

Appendix F. The Test-Curriculum Matching Analysis Science TIMSS 2011 INTERNATIONAL RESULTS IN SCIENCE APPENDIX F 479

Appendix F. The Test-Curriculum Matching Analysis Science TIMSS 2011 INTERNATIONAL RESULTS IN SCIENCE APPENDIX F 479 Appendix F The Test-Curriculum Matching Analysis Science TIMSS 2011 INTERNATIONAL RESULTS IN SCIENCE APPENDIX F 479 TIMSS went to great lengths to ensure that comparisons of student achievement across

More information

COMPETITIVENESS OF CANADIAN AGRI-FOOD EXPORTS

COMPETITIVENESS OF CANADIAN AGRI-FOOD EXPORTS JOINT SERIES OF COMPETITIVENESS NUMBER 18 NOVEMBER 1999 COMPETITIVENESS OF CANADIAN AGRI-FOOD EXPORTS AGAINST ITS COMPETITORS' IN ASIA: 198-97 Kevin Chen and Yufeng Duan Department of Rutal Economy University

More information

Development of novel algorithm by combining Wavelet based Enhanced Canny edge Detection and Adaptive Filtering Method for Human Emotion Recognition

Development of novel algorithm by combining Wavelet based Enhanced Canny edge Detection and Adaptive Filtering Method for Human Emotion Recognition International Journal of Engineering Research and Development e-issn: 2278-067X, p-issn: 2278-800X, www.ijerd.com Volume 12, Issue 9 (September 2016), PP.67-72 Development of novel algorithm by combining

More information

Global Coverage. Regional Coverage. Country Coverage Company Coverage

Global Coverage. Regional Coverage. Country Coverage Company Coverage Global Fatty Liver Disease Drugs Market By Type (ALD, NAFLD), By Stages of ALD and NAFLD, By Region, By Country: Opportunities and Forecast (2017-2022) By Type- ALD and NAFLD ALD Stages - Alcoholic Steatosis

More information

School Curriculum and Standards Authority

School Curriculum and Standards Authority Accounting and Finance Stage 2-74 42 12 57 9 43 21-50 Accounting and Finance Stage 3-1366 1181 596 48 635 52 1231 4.2 Ancient History Stage 2-4 10 2 40 3 60 5-50 Ancient History Stage 3-200 212 92 60 62

More information

ESOL Skills for Life Entry 3 Reading

ESOL Skills for Life Entry 3 Reading ESOL Skills for Life Entry 3 Reading Sample paper 5 Time allowed: 60 minutes Please answer all questions. Circle your answers in pen, not pencil, on the separate answer sheet. You may not use dictionaries.

More information

THE GLOBAL HEALTH SECURITY AGENDA. We have to change our mindsets and start thinking about biological threats as the security threats that they are.

THE GLOBAL HEALTH SECURITY AGENDA. We have to change our mindsets and start thinking about biological threats as the security threats that they are. THE GLOBAL HEALTH SECURITY AGENDA We have to change our mindsets and start thinking about biological threats as the security threats that they are. President Barack Obama September 26, 2014 1 Global Health

More information

Segmentation of Cell Membrane and Nucleus by Improving Pix2pix

Segmentation of Cell Membrane and Nucleus by Improving Pix2pix Segmentation of Membrane and Nucleus by Improving Pix2pix Masaya Sato 1, Kazuhiro Hotta 1, Ayako Imanishi 2, Michiyuki Matsuda 2 and Kenta Terai 2 1 Meijo University, Siogamaguchi, Nagoya, Aichi, Japan

More information

Data Visualization - Basics

Data Visualization - Basics Data Visualization - Basics The Visualization Process ACQUIRE PARSE FILTER MINE REPRESENT REFINE INTERACT Sales Data (US $ in thousands) Region Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Domestic

More information

B657: Final Project Report Holistically-Nested Edge Detection

B657: Final Project Report Holistically-Nested Edge Detection B657: Final roject Report Holistically-Nested Edge Detection Mingze Xu & Hanfei Mei May 4, 2016 Abstract Holistically-Nested Edge Detection (HED), which is a novel edge detection method based on fully

More information

Emotion Recognition using a Cauchy Naive Bayes Classifier

Emotion Recognition using a Cauchy Naive Bayes Classifier Emotion Recognition using a Cauchy Naive Bayes Classifier Abstract Recognizing human facial expression and emotion by computer is an interesting and challenging problem. In this paper we propose a method

More information

More information at

More information at Report Information More information at https://www.htfmarketreport.com/reports/1288313 Global Homeopathic Products Market Professional Survey Report 2018 Report Code: HTF1288313 Pages: 111 Price: 1-User

More information

Your Community in Profile: Halton-Peel

Your Community in Profile: Halton-Peel Your Community in Profile: Halton-Peel Peel Halton Building healthy and vibrant communities The Ontario Trillium Foundation is an agency of the Government of Ontario. Table of Contents Introduction...4

More information

TWO HANDED SIGN LANGUAGE RECOGNITION SYSTEM USING IMAGE PROCESSING

TWO HANDED SIGN LANGUAGE RECOGNITION SYSTEM USING IMAGE PROCESSING 134 TWO HANDED SIGN LANGUAGE RECOGNITION SYSTEM USING IMAGE PROCESSING H.F.S.M.Fonseka 1, J.T.Jonathan 2, P.Sabeshan 3 and M.B.Dissanayaka 4 1 Department of Electrical And Electronic Engineering, Faculty

More information

Use and Perception of Scientific Medical Reprints

Use and Perception of Scientific Medical Reprints TITLE OF PRESENTATION Use and Perception of Scientific Medical Reprints Elsevier Health Panel Research Report Date August 2017 I 2 Report Structure Research Objectives, Methodology & Sample Executive Summary

More information

Global Terephthalic Acid (PTA) Market Study ( )

Global Terephthalic Acid (PTA) Market Study ( ) Global Terephthalic Acid (PTA) Market Study (2014 2025) Table of Contents 1. INTRODUCTION 1.1. Introduction to Terephthalic Acid Market Product Description Properties Industry Structure Value Chain Market

More information

Short Range Outlook for Steel Demand Autumn OECD Steel Committee, Paris, 10 December Short Range Outlook overview

Short Range Outlook for Steel Demand Autumn OECD Steel Committee, Paris, 10 December Short Range Outlook overview Short Range Outlook for Steel Demand Autumn 29 OECD Steel Committee, Paris, 1 December 29 Content Short Range Outlook overview Background of the Forecast Regional overviews Conclusion 2 1 Short Range Outlook

More information

Highly Accurate Brain Stroke Diagnostic System and Generative Lesion Model. Junghwan Cho, Ph.D. CAIDE Systems, Inc. Deep Learning R&D Team

Highly Accurate Brain Stroke Diagnostic System and Generative Lesion Model. Junghwan Cho, Ph.D. CAIDE Systems, Inc. Deep Learning R&D Team Highly Accurate Brain Stroke Diagnostic System and Generative Lesion Model Junghwan Cho, Ph.D. CAIDE Systems, Inc. Deep Learning R&D Team Established in September, 2016 at 110 Canal st. Lowell, MA 01852,

More information

arxiv: v2 [cs.cv] 22 Mar 2018

arxiv: v2 [cs.cv] 22 Mar 2018 Deep saliency: What is learnt by a deep network about saliency? Sen He 1 Nicolas Pugeault 1 arxiv:1801.04261v2 [cs.cv] 22 Mar 2018 Abstract Deep convolutional neural networks have achieved impressive performance

More information

Global E-Cigarette & Vaporizer Market Research Report - Forecast to 2027

Global E-Cigarette & Vaporizer Market Research Report - Forecast to 2027 Report Information More information from: https://www.marketresearchfuture.com/reports/788 Global E-Cigarette & Vaporizer Market Research Report - Forecast to 2027 Report / Search Code: MRFR/CR/0304-CRR

More information

CONNECTED HEALTH TRENDS

CONNECTED HEALTH TRENDS CONNECTED HEALTH TRENDS 2018 A CONNECTED HEALTH MOVING IN THE RIGHT DIRECTION The figure 10,000 is now synonymous with the number of steps we should all be striving for each day to help maintain a healthy

More information

Automatic Classification of Perceived Gender from Facial Images

Automatic Classification of Perceived Gender from Facial Images Automatic Classification of Perceived Gender from Facial Images Joseph Lemley, Sami Abdul-Wahid, Dipayan Banik Advisor: Dr. Razvan Andonie SOURCE 2016 Outline 1 Introduction 2 Faces - Background 3 Faces

More information

OIE Situation Report for Avian Influenza

OIE Situation Report for Avian Influenza OIE Situation Report for Avian Influenza Latest update: 25/01/2018 The epidemiology of avian influenza is complex. The virus constantly evolves and the behavior of each new subtype (and strains within

More information

The Self Care: Be Your Best Report A global study on common health conditions and how people practice self care

The Self Care: Be Your Best Report A global study on common health conditions and how people practice self care The Self Care: Be Your Best Report 2018 A global study on common health conditions and how people practice self care Foreword Most of us have woken up recently with some type of pain, a blocked or runny

More information

Detection of suspicious lesion based on Multiresolution Analysis using windowing and adaptive thresholding method.

Detection of suspicious lesion based on Multiresolution Analysis using windowing and adaptive thresholding method. Detection of suspicious lesion based on Multiresolution Analysis using windowing and adaptive thresholding method. Ms. N. S. Pande Assistant Professor, Department of Computer Science and Engineering,MGM

More information

Facial Expression Classification Using Convolutional Neural Network and Support Vector Machine

Facial Expression Classification Using Convolutional Neural Network and Support Vector Machine Facial Expression Classification Using Convolutional Neural Network and Support Vector Machine Valfredo Pilla Jr, André Zanellato, Cristian Bortolini, Humberto R. Gamba and Gustavo Benvenutti Borba Graduate

More information

POST SHOW REPORT. The Leading Dental Conference & Exhibition for Middle-East, Africa and South Asia markets.

POST SHOW REPORT. The Leading Dental Conference & Exhibition for Middle-East, Africa and South Asia markets. The Leading Dental Conference & Exhibition for Middle-East, Africa and South Asia markets. 22 Editions of Success to Reach Excellence. Organised by Strategic Partner POST SHOW REPORT 2018 Supported by

More information

Influenza Vaccination Market Research Report Global Forecast till 2023

Influenza Vaccination Market Research Report Global Forecast till 2023 Report Information More information from: https://www.marketresearchfuture.com/reports/1621 Influenza Vaccination Market Research Report Global Forecast till 2023 Report / Search Code: MRFR/HC/1090-HCRR

More information

NMF-Density: NMF-Based Breast Density Classifier

NMF-Density: NMF-Based Breast Density Classifier NMF-Density: NMF-Based Breast Density Classifier Lahouari Ghouti and Abdullah H. Owaidh King Fahd University of Petroleum and Minerals - Department of Information and Computer Science. KFUPM Box 1128.

More information

FEATURE EXTRACTION USING GAZE OF PARTICIPANTS FOR CLASSIFYING GENDER OF PEDESTRIANS IN IMAGES

FEATURE EXTRACTION USING GAZE OF PARTICIPANTS FOR CLASSIFYING GENDER OF PEDESTRIANS IN IMAGES FEATURE EXTRACTION USING GAZE OF PARTICIPANTS FOR CLASSIFYING GENDER OF PEDESTRIANS IN IMAGES Riku Matsumoto, Hiroki Yoshimura, Masashi Nishiyama, and Yoshio Iwai Department of Information and Electronics,

More information

This is a repository copy of Measuring the effect of public health campaigns on Twitter: the case of World Autism Awareness Day.

This is a repository copy of Measuring the effect of public health campaigns on Twitter: the case of World Autism Awareness Day. This is a repository copy of Measuring the effect of public health campaigns on Twitter: the case of World Autism Awareness Day. White Rose Research Online URL for this paper: http://eprints.whiterose.ac.uk/127215/

More information

arxiv: v3 [stat.ml] 27 Mar 2018

arxiv: v3 [stat.ml] 27 Mar 2018 ATTACKING THE MADRY DEFENSE MODEL WITH L 1 -BASED ADVERSARIAL EXAMPLES Yash Sharma 1 and Pin-Yu Chen 2 1 The Cooper Union, New York, NY 10003, USA 2 IBM Research, Yorktown Heights, NY 10598, USA sharma2@cooper.edu,

More information

CYCLOTRONS WORLD MARKET REPORT & DIRECTORY EDITION TOC and Summary USED IN NUCLEAR MEDICINE MARKET DATA, COMPANIES PROFILES.

CYCLOTRONS WORLD MARKET REPORT & DIRECTORY EDITION TOC and Summary USED IN NUCLEAR MEDICINE MARKET DATA, COMPANIES PROFILES. CYCLOTRONS USED IN NUCLEAR MEDICINE WORLD MARKET REPORT & DIRECTORY MARKET DATA, COMPANIES PROFILES TOC and Summary EDITION 2015 A report written by Paul-Emmanuel Goethals & Richard Zimmermann Page 1/5

More information

Learning Multifunctional Binary Codes for Both Category and Attribute Oriented Retrieval Tasks

Learning Multifunctional Binary Codes for Both Category and Attribute Oriented Retrieval Tasks Learning Multifunctional Binary Codes for Both Category and Attribute Oriented Retrieval Tasks Haomiao Liu 1,2, Ruiping Wang 1,2,3, Shiguang Shan 1,2,3, Xilin Chen 1,2,3 1 Key Laboratory of Intelligent

More information

Prentice Hall. World Civilizations, The Global Experience, 4th Edition, AP* Edition Grades 9-12

Prentice Hall. World Civilizations, The Global Experience, 4th Edition, AP* Edition Grades 9-12 Prentice Hall Grades 9-12 World Civilizations, The Global Experience, 4th Edition, AP* Edition 2006 C O R R E L A T E D T O Grades 9-12 FOUNDATIONS: C.8000 B.C.E. TO 600 C.E. What students are expected

More information

Global Continuous Positive Airway Pressure(CPAP) Devices Market Research Report - Forecast to 2027

Global Continuous Positive Airway Pressure(CPAP) Devices Market Research Report - Forecast to 2027 Report Information More information from: https://www.marketresearchfuture.com/reports/939 Global Continuous Positive Airway Pressure(CPAP) Devices Market Research Report - Forecast to 2027 Report / Search

More information

Deep Learning based FACS Action Unit Occurrence and Intensity Estimation

Deep Learning based FACS Action Unit Occurrence and Intensity Estimation Deep Learning based FACS Action Unit Occurrence and Intensity Estimation Amogh Gudi, H. Emrah Tasli, Tim M. den Uyl, Andreas Maroulis Vicarious Perception Technologies, Amsterdam, The Netherlands Abstract

More information

Dual Path Network and Its Applications

Dual Path Network and Its Applications Learning and Vision Group (NUS), ILSVRC 2017 - CLS-LOC & DET tasks Dual Path Network and Its Applications National University of Singapore: Yunpeng Chen, Jianan Li, Huaxin Xiao, Jianshu Li, Xuecheng Nie,

More information

One intelligent solution

One intelligent solution One intelligent solution System One sleep therapy platform gets even smarter with advanced intelligence for optimum care and easier patient management. Industry-leading enhancements and innovations The

More information

What will women s empowerment mean for men? Selected facts and figures to spark your thinking

What will women s empowerment mean for men? Selected facts and figures to spark your thinking What will women s empowerment mean for men? Selected facts and figures to spark your thinking Prepared for the Women s Forum for the Economy and Society by PwC October 211 What will women s empowerment

More information

Face Analysis : Identity vs. Expressions

Face Analysis : Identity vs. Expressions Hugo Mercier, 1,2 Patrice Dalle 1 Face Analysis : Identity vs. Expressions 1 IRIT - Université Paul Sabatier 118 Route de Narbonne, F-31062 Toulouse Cedex 9, France 2 Websourd Bâtiment A 99, route d'espagne

More information

Summary. About Pilot Waverly Labs is... Core Team Pilot Translation Kit FAQ Pilot Speech Translation App FAQ Testimonials Contact Useful Links. p.

Summary. About Pilot Waverly Labs is... Core Team Pilot Translation Kit FAQ Pilot Speech Translation App FAQ Testimonials Contact Useful Links. p. Press kit Summary About Pilot Waverly Labs is... Core Team Pilot Translation Kit FAQ Pilot Speech Translation App FAQ Testimonials Contact Useful Links p. 3 p.4 p. 5 p. 6 p. 8 p. 10 p. 12 p. 14 p. 16 p.

More information

Classification of Smoking Status: The Case of Turkey

Classification of Smoking Status: The Case of Turkey Classification of Smoking Status: The Case of Turkey Zeynep D. U. Durmuşoğlu Department of Industrial Engineering Gaziantep University Gaziantep, Turkey unutmaz@gantep.edu.tr Pınar Kocabey Çiftçi Department

More information

WAVERLY LABS - Press Kit

WAVERLY LABS - Press Kit Press kit Summary About Pilot Waverly Labs Core Team Pilot Translating Earpiece FAQ Pilot Speech Translator App FAQ Testimonials Contact Useful Links p. 3 p.4 p. 5 p. 6 p. 8 p. 10 p. 12 p. 14 p. 16 p.

More information

Global Polybutylene Terephthalate (PBT) Market Study ( )

Global Polybutylene Terephthalate (PBT) Market Study ( ) Global Polybutylene Terephthalate (PBT) Market Study (2014 2025) Table of Contents 1. INTRODUCTION 1.1. Introduction to Polybutylene Terephthalate Market Product Description Properties Industry Structure

More information

Efficient Deep Model Selection

Efficient Deep Model Selection Efficient Deep Model Selection Jose Alvarez Researcher Data61, CSIRO, Australia GTC, May 9 th 2017 www.josemalvarez.net conv1 conv2 conv3 conv4 conv5 conv6 conv7 conv8 softmax prediction???????? Num Classes

More information

Image-Based Estimation of Real Food Size for Accurate Food Calorie Estimation

Image-Based Estimation of Real Food Size for Accurate Food Calorie Estimation Image-Based Estimation of Real Food Size for Accurate Food Calorie Estimation Takumi Ege, Yoshikazu Ando, Ryosuke Tanno, Wataru Shimoda and Keiji Yanai Department of Informatics, The University of Electro-Communications,

More information

Healthcare Research You

Healthcare Research You DL for Healthcare Goals Healthcare Research You What are high impact problems in healthcare that deep learning can solve? What does research in AI applications to medical imaging look like? How can you

More information

Differentiating Tumor and Edema in Brain Magnetic Resonance Images Using a Convolutional Neural Network

Differentiating Tumor and Edema in Brain Magnetic Resonance Images Using a Convolutional Neural Network Original Article Differentiating Tumor and Edema in Brain Magnetic Resonance Images Using a Convolutional Neural Network Aida Allahverdi 1, Siavash Akbarzadeh 1, Alireza Khorrami Moghaddam 2, Armin Allahverdy

More information

Towards Open Set Deep Networks: Supplemental

Towards Open Set Deep Networks: Supplemental Towards Open Set Deep Networks: Supplemental Abhijit Bendale*, Terrance E. Boult University of Colorado at Colorado Springs {abendale,tboult}@vast.uccs.edu In this supplement, we provide we provide additional

More information

Virtual reproduction of the migration flows generated by AIESEC

Virtual reproduction of the migration flows generated by AIESEC Virtual reproduction of the migration flows generated by AIESEC P.G. Battaglia, F.Gorrieri, D.Scorpiniti Introduction AIESEC is an international non- profit organization that provides services for university

More information

An Overview and Comparative Analysis on Major Generative Models

An Overview and Comparative Analysis on Major Generative Models An Overview and Comparative Analysis on Major Generative Models Zijing Gu zig021@ucsd.edu Abstract The amount of researches on generative models has been grown rapidly after a period of silence due to

More information

Lactase Investigation

Lactase Investigation Lactase Investigation Introduction Name: Lactose is a type of sugar in milk. Other than milk and some dairy products made from milk, lactose is not found naturally in foods humans eat. Lactose is a type

More information

Recommending Outfits from Personal Closet

Recommending Outfits from Personal Closet Recommending Outfits from Personal Closet Pongsate Tangseng 1, Kota Yamaguchi 2, and Takayuki Okatani 1,3 1 Tohoku University, Sendai, Japan 2 CyberAgent, Inc., Tokyo, Japan 3 RIKEN Center for AIP, Tokyo,

More information

Image Representations and New Domains in Neural Image Captioning. Jack Hessel, Nic Saava, Mike Wilber

Image Representations and New Domains in Neural Image Captioning. Jack Hessel, Nic Saava, Mike Wilber Image Representations and New Domains in Neural Image Captioning Jack Hessel, Nic Saava, Mike Wilber 1 The caption generation problem...? A young girl runs through a grassy field. 1? A you throug 1 Image

More information

Tobacco Insights May

Tobacco Insights May Tobacco Insights 2015 - May Product Code: TI0017MR Published Date: June 2015 Report Price: US$ 150 (Single Copy) Synopsis Tobacco Insights 2015 - May report is a qualitative report providing extensive

More information

An Artificial Neural Network Architecture Based on Context Transformations in Cortical Minicolumns

An Artificial Neural Network Architecture Based on Context Transformations in Cortical Minicolumns An Artificial Neural Network Architecture Based on Context Transformations in Cortical Minicolumns 1. Introduction Vasily Morzhakov, Alexey Redozubov morzhakovva@gmail.com, galdrd@gmail.com Abstract Cortical

More information

PathGAN: Visual Scanpath Prediction with Generative Adversarial Networks

PathGAN: Visual Scanpath Prediction with Generative Adversarial Networks PathGAN: Visual Scanpath Prediction with Generative Adversarial Networks Marc Assens 1, Kevin McGuinness 1, Xavier Giro-i-Nieto 2, and Noel E. O Connor 1 1 Insight Centre for Data Analytic, Dublin City

More information

Viewpoint Invariant Convolutional Networks for Identifying Risky Hand Hygiene Scenarios

Viewpoint Invariant Convolutional Networks for Identifying Risky Hand Hygiene Scenarios Viewpoint Invariant Convolutional Networks for Identifying Risky Hand Hygiene Scenarios Michelle Guo 1 Albert Haque 1 Serena Yeung 1 Jeffrey Jopling 1 Lance Downing 1 Alexandre Alahi 1,2 Brandi Campbell

More information

Sleep apnea devices Market Research Report- Global Forecast To 2022

Sleep apnea devices Market Research Report- Global Forecast To 2022 Report Information More information from: https://www.marketresearchfuture.com/reports/2061 Sleep apnea devices Market Research Report- Global Forecast To 2022 Report / Search Code: MRFR/HC/1529-HCRR Publish

More information

Global Cosmetic Dentistry Market Research Report 2018

Global Cosmetic Dentistry Market Research Report 2018 Report Information More information from: https://www.wiseguyreports.com/reports/3029111-global-cosmetic-dentistry-market-research-report- 2018 Global Cosmetic Dentistry Market Research Report 2018 Report

More information

CHAPTER I. CONTENTS.

CHAPTER I. CONTENTS. 199 CHAPTER I. CONTENTS. 200 201 202 THE INCIDENCE OF EPIDEMIC INFLUENZA DURING 1918-19 IN EUROPE AND IN THE WESTERN HEMISPHERE. BY R. BRUCE Low, C.B., M.D. INTRODUCTION. When it was first recognised that

More information

Hypothesis-Driven Research

Hypothesis-Driven Research Hypothesis-Driven Research Research types Descriptive science: observe, describe and categorize the facts Discovery science: measure variables to decide general patterns based on inductive reasoning Hypothesis-driven

More information

Proton Therapy Market Outlook - Global Analysis

Proton Therapy Market Outlook - Global Analysis Proton Therapy Market Outlook - Global Analysis Proton Therapy Market Outlook - Global Analysis BioPortfolio has been marketing business and market research reports from selected publishers for over fifteen

More information

OIE Situation Report for Highly Pathogenic Avian Influenza

OIE Situation Report for Highly Pathogenic Avian Influenza OIE Situation Report for Highly Pathogenic Avian Influenza Latest update: 28/02/2018 The epidemiology of avian influenza is complex. The virus constantly evolves and the behavior of each new subtype (and

More information

OIE Situation Report for Highly Pathogenic Avian Influenza

OIE Situation Report for Highly Pathogenic Avian Influenza OIE Situation Report for Highly Pathogenic Avian Influenza Latest update: 31/05/2018 The epidemiology of avian influenza (AI) is complex. The AI virus constantly evolves by mutation and re-assortment with

More information