Ji Young Lee, Franck Dernoncourt, Peter Szolovits
Comments: The first two authors contributed equally to this work
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE); Machine Learning (stat.ML)
Recent approaches based on artificial neural networks (ANNs) have shown
promising results for named-entity recognition (NER). In order to achieve high
performances, ANNs need to be trained on a large labeled dataset. However,
labels might be difficult to obtain for the dataset on which the user wants to
perform NER: label scarcity is particularly pronounced for patient note
de-identification, which is an instance of NER. In this work, we analyze to
what extent transfer learning may address this issue. In particular, we
demonstrate that transferring an ANN model trained on a large labeled dataset
to another dataset with a limited number of labels improves upon the
state-of-the-art results on two different datasets for patient note
de-identification.
Ankur Sinha, Pekka Malo, Kalyanmoy Deb
Subjects: Optimization and Control (math.OC); Neural and Evolutionary Computing (cs.NE)
Bilevel optimization is defined as a mathematical program, where an
optimization problem contains another optimization problem as a constraint.
These problems have received significant attention from the mathematical
programming community. Only limited work exists on bilevel problems using
evolutionary computation techniques; however, recently there has been an
increasing interest due to the proliferation of practical applications and the
potential of evolutionary algorithms in tackling these problems. This paper
provides a comprehensive review on bilevel optimization from the basic
principles to solution strategies; both classical and evolutionary. A number of
potential application problems are also discussed. To offer the readers
insights on the prominent developments in the field of bilevel optimization, we
have performed an automated text-analysis of an extended list of papers
published on bilevel optimization to date. This paper should motivate
evolutionary computation researchers to pay more attention to this practical
yet challenging area.
Nicolas Audebert (Palaiseau, OBELIX), Bertrand Le Saux (Palaiseau), Sébastien Lefèvre (OBELIX)
Journal-ref: EARTHVISION 2017 IEEE/ISPRS CVPR Workshop. Large Scale Computer
Vision for Remote Sensing Imagery, Jul 2017, Honolulu, United States. 2017
Subjects: Computer Vision and Pattern Recognition (cs.CV); Neural and Evolutionary Computing (cs.NE)
In this work, we investigate the use of OpenStreetMap data for semantic
labeling of Earth Observation images. Deep neural networks have been used in
the past for remote sensing data classification from various sensors, including
multispectral, hyperspectral, SAR and LiDAR data. While OpenStreetMap has
already been used as ground truth data for training such networks, this
abundant data source remains rarely exploited as an input information layer. In
this paper, we study different use cases and deep network architectures to
leverage OpenStreetMap data for semantic labeling of aerial and satellite
images. Especially , we look into fusion based architectures and coarse-to-fine
segmentation to include the OpenStreetMap layer into multispectral-based deep
fully convolutional networks. We illustrate how these methods can be
successfully used on two public datasets: ISPRS Potsdam and DFC2017. We show
that OpenStreetMap data can efficiently be integrated into the vision-based
deep learning models and that it significantly improves both the accuracy
performance and the convergence speed of the networks.
Qingxing Dong, Xin Zhou
Comments: 24 pages, 3 figures
Subjects: Social and Information Networks (cs.SI); Neural and Evolutionary Computing (cs.NE)
Opinion polarization is a ubiquitous phenomenon in opinion dynamics. In
contrast to the traditional consensus oriented group decision making (GDM)
framework, this paper proposes a framework with the co-evolution of both
opinions and relationship networks to improve the potential consensus level of
a group and help the group reach a stable state. Taking the bound of confidence
and the degree of individual’s persistence into consideration, the evolution of
the opinion is driven by the relationship among the group. Meanwhile, the
antagonism or cooperation of individuals presented by the network topology also
evolve according to the dynamic opinion distances. Opinions are convergent and
the stable state will be reached in this co-evolution mechanism. We further
explored this framework through simulation experiments. The simulation results
verify the influence of the level of persistence on the time cost and indicate
the influence of group size, the initial topology of networks and the bound of
confidence on the number of opinion clusters.
Stanislav Filippov, Arsenii Moiseev, Andronenko Andrey
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Coronary angiography is considered to be a safe tool for the evaluation of
coronary artery disease and perform in approximately 12 million patients each
year worldwide. In most cases, angiograms are manually analyzed by a
cardiologist. Actually, there are no clinical practice algorithms which could
improve and automate this work. Neural networks show high efficiency in tasks
of image analysis and they can be used for the analysis of angiograms and
facilitate diagnostics. We have developed an algorithm based on Convolutional
Neural Network and Neural Network U-Net for vessels segmentation and defects
detection such as stenosis. For our research we used anonymized angiography
data obtained from one of the city hospitals and augmented them to improve
learning efficiency. U-Net usage provided high quality segmentation and the
combination of our algorithm with an ensemble of classifiers shows a good
accuracy in the task of ischemia evaluation on test data. Subsequently, this
approach can be served as a basis for the creation of an analytical system that
could speed up the diagnosis of cardiovascular diseases and greatly facilitate
the work of a specialist.
Min Tang, Sepehr Valipour, Zichen Vincent Zhang, Dana Cobzas, MartinJagersand
Subjects: Computer Vision and Pattern Recognition (cs.CV)
This paper proposes a novel image segmentation approachthat integrates fully
convolutional networks (FCNs) with a level setmodel. Compared with a FCN, the
integrated method can incorporatesmoothing and prior information to achieve an
accurate segmentation.Furthermore, different than using the level set model as
a post-processingtool, we integrate it into the training phase to fine-tune the
FCN. Thisallows the use of unlabeled data during training in a
semi-supervisedsetting. Using two types of medical imaging data (liver CT and
left ven-tricle MRI data), we show that the integrated method achieves
goodperformance even when little training data is available, outperformingthe
FCN or the level set model alone.
Mehmet Turan, Abdullah Abdullah, Redhwan Jamiruddin, Helder Araujo, Ender Konukoglu, Metin Sitti
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Since its development, ingestible wireless endoscopy is considered to be a
painless diagnostic method to detect a number of diseases inside GI tract.
Medical related engineering companies have made significant improvements in
this technology in last decade; however, some major limitations still residue.
Localization of the next generation steerable endoscopic capsule robot in
six-degree-of freedom (6 DoF) and active motion control are some of these
limitations. The significance of localization capability concerns with the
correct diagnosis of the disease area. This paper presents a very robust 6-DoF
localization method based on supervised training of an architecture consisting
of recurrent networks (RNN) embedded into a convolutional neural network (CNN)
to make use of both just-in-moment information obtained by CNN and correlative
information across frames obtained by RNN. To our knowledge, the idea of
embedding RNNs into a CNN architecture is for the first time proposed in
literature. The experimental results show that the proposed RNN-in-CNN
architecture performs very well for endoscopic capsule robot localization in
cases reflection distortions, noise, sudden camera movements and lack of
distinguishable features.
Mairéad Grogan, Rozenn Dahyot
Subjects: Computer Vision and Pattern Recognition (cs.CV)
We present a flexible approach to colour transfer inspired by techniques
recently proposed for shape registration. Colour distributions of the palette
and target images are modelled with Gaussian Mixture Models (GMMs) that are
robustly registered to infer a non linear parametric transfer function. We show
experimentally that our approach compares well to current techniques both
quantitatively and qualitatively. Moreover, our technique is computationally
the fastest and can take efficient advantage of parallel processing
architectures for recolouring images and videos. Our transfer function is
parametric and hence can be stored in memory for later usage and also combined
with other computed transfer functions to create interesting visual effects.
Overall this paper provides a fast user friendly approach to recolouring of
image and video materials.
Nicolas Audebert (Palaiseau, OBELIX), Bertrand Le Saux (Palaiseau), Sébastien Lefèvre (OBELIX)
Journal-ref: EARTHVISION 2017 IEEE/ISPRS CVPR Workshop. Large Scale Computer
Vision for Remote Sensing Imagery, Jul 2017, Honolulu, United States. 2017
Subjects: Computer Vision and Pattern Recognition (cs.CV); Neural and Evolutionary Computing (cs.NE)
In this work, we investigate the use of OpenStreetMap data for semantic
labeling of Earth Observation images. Deep neural networks have been used in
the past for remote sensing data classification from various sensors, including
multispectral, hyperspectral, SAR and LiDAR data. While OpenStreetMap has
already been used as ground truth data for training such networks, this
abundant data source remains rarely exploited as an input information layer. In
this paper, we study different use cases and deep network architectures to
leverage OpenStreetMap data for semantic labeling of aerial and satellite
images. Especially , we look into fusion based architectures and coarse-to-fine
segmentation to include the OpenStreetMap layer into multispectral-based deep
fully convolutional networks. We illustrate how these methods can be
successfully used on two public datasets: ISPRS Potsdam and DFC2017. We show
that OpenStreetMap data can efficiently be integrated into the vision-based
deep learning models and that it significantly improves both the accuracy
performance and the convergence speed of the networks.
Yeong-Jun Cho, Kuk-Jin Yoon
Comments: 12 pages, 12 figures, 4 tables
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Person re-identification is the problem of recognizing people across
different images or videos with non-overlapping views. Although there has been
much progress in person re-identification over the last decade, it remains a
challenging task because appearances of people can seem extremely different
across diverse camera viewpoints and person poses. In this paper, we propose a
novel framework for person re-identification by analyzing camera viewpoints and
person poses in a so-called Pose-aware Multi-shot Matching (PaMM), which
robustly estimates people’s poses and efficiently conducts multi-shot matching
based on pose information. Experimental results using public person
re-identification datasets show that the proposed methods outperform
state-of-the-art methods and are promising for person re-identification from
diverse viewpoints and pose variances.
Abhishek Sharma
Comments: 8 pages, Under Review
Subjects: Computer Vision and Pattern Recognition (cs.CV); Learning (cs.LG)
This paper presents a novel framework in which image cosegmentation and
colocalization are cast into a single optimization problem that integrates
information from low level appearance cues with that of high level localization
cues in a very weakly supervised manner. In contrast to multi-task learning
paradigm that learns similar tasks using a shared representation, the proposed
framework leverages two representations at different levels and simultaneously
discriminates between foreground and background at the bounding box and
superpixel level using discriminative clustering. We show empirically that
constraining the two problems at different scales enables the transfer of
semantic localization cues to improve cosegmentation output whereas local
appearance based segmentation cues help colocalization. The unified framework
outperforms strong baseline approaches, of learning the two problems
separately, by a large margin on four benchmark datasets. Furthermore, it
obtains competitive results compared to the state of the art for cosegmentation
on two benchmark datasets and second best result for colocalization on Pascal
VOC 2007.
Dong Yang, Tao Xiong, Daguang Xu, Qiangui Huang, David Liu, S.Kevin Zhou, Zhoubing Xu, JinHyeong Park, Mingqing Chen, Trac D. Tran, Sang Peter Chin, Dimitris Metaxas, Dorin Comaniciu
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Automatic localization and labeling of vertebra in 3D medical images plays an
important role in many clinical tasks, including pathological diagnosis,
surgical planning and postoperative assessment. However, the unusual conditions
of pathological cases, such as the abnormal spine curvature, bright visual
imaging artifacts caused by metal implants, and the limited field of view,
increase the difficulties of accurate localization. In this paper, we propose
an automatic and fast algorithm to localize and label the vertebra centroids in
3D CT volumes. First, we deploy a deep image-to-image network (DI2IN) to
initialize vertebra locations, employing the convolutional encoder-decoder
architecture together with multi-level feature concatenation and deep
supervision. Next, the centroid probability maps from DI2IN are iteratively
evolved with the message passing schemes based on the mutual relation of
vertebra centroids. Finally, the localization results are refined with sparsity
regularization. The proposed method is evaluated on a public dataset of 302
spine CT volumes with various pathologies. Our method outperforms other
state-of-the-art methods in terms of localization accuracy. The run time is
around 3 seconds on average per case. To further boost the performance, we
retrain the DI2IN on additional 1000+ 3D CT volumes from different patients. To
the best of our knowledge, this is the first time more than 1000 3D CT volumes
with expert annotation are adopted in experiments for the anatomic landmark
detection tasks. Our experimental results show that training with such a large
dataset significantly improves the performance and the overall identification
rate, for the first time by our knowledge, reaches 90 %.
Shikun Liu, Alexander G. Ororbia II, C. Lee Giles
Comments: 10 pages, 8 figures, 2 tables
Subjects: Computer Vision and Pattern Recognition (cs.CV)
We propose the Variational Shape Learner (VSL), a hierarchical
latent-variable model for 3D shape learning. VSL employs an unsupervised,
variational approach to the inference and learning of the underlying structure
of voxelized 3D shapes. Our model successfully learns 3D shapes via a
hierarchical latent representation, made possible through the use of
skip-connections. Realistic 3D objects can be generated by sampling its latent
probabilistic manifold. We show that our inference and generative models can be
trained end-to-end from 2D images to perform single image 3D model retrieval.
Experiments show the improved performance of our model both quantitatively and
qualitatively over a range of tasks.
Subarna Tripathi, Gokce Dane, Byeongkeun Kang, Vasudev Bhaskaran, Truong Nguyen
Comments: Embedded Vision Workshop in CVPR
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Deep convolutional Neural Networks (CNN) are the state-of-the-art performers
for object detection task. It is well known that object detection requires more
computation and memory than image classification. Thus the consolidation of a
CNN-based object detection for an embedded system is more challenging. In this
work, we propose LCDet, a fully-convolutional neural network for generic object
detection that aims to work in embedded systems. We design and develop an
end-to-end TensorFlow(TF)-based model. Additionally, we employ 8-bit
quantization on the learned weights. We use face detection as a use case. Our
TF-Slim based network can predict different faces of different shapes and sizes
in a single forward pass. Our experimental results show that the proposed
method achieves comparative accuracy comparing with state-of-the-art CNN-based
face detection methods, while reducing the model size by 3x and memory-BW by
~4x comparing with one of the best real-time CNN-based object detector such as
YOLO. TF 8-bit quantized model provides additional 4x memory reduction while
keeping the accuracy as good as the floating point model. The proposed model
thus becomes amenable for embedded implementations.
Benjamin S. Kunsberg, Daniel Niels Holtmann-Rice, Steven W. Zucker
Subjects: Computer Vision and Pattern Recognition (cs.CV)
We continue the development of a linear algebraic framework for the
shape-from-shading problem, exploiting the manner in which tensors arise when
scalar (e.g. image) and vector (e.g. surface normal) fields are differentiated
multiple times. In this paper we apply that framework to develop Taylor
expansions of the normal field and build a boot-strapping algorithm to find
these polynomial surface solutions (under any light source) consistent with a
given patch to arbitrary order. A generic constraint on the image derivatives
restricts these solutions to a 2-D subspace, plus an unknown rotation matrix.
The parameters for the subspace and rotation matrix encapsulate the ambiguity
in the shading problem.
Daniel Niels Holtmann-Rice, Benjamin S. Kunsberg, Steven W. Zucker
Subjects: Computer Vision and Pattern Recognition (cs.CV)
We develop a linear algebraic framework for the shape-from-shading problem,
because tensors arise when scalar (e.g. image) and vector (e.g. surface normal)
fields are differentiated multiple times. The work is in two parts. In this
first part we investigate when image derivatives exhibit invariance to changing
illumination by calculating the statistics of image derivatives under general
distributions on the light source. We computationally validate the hypothesis
that image orientations (derivatives) provide increased invariance to
illumination by showing (for a Lambertian model) that a shape-from-shading
algorithm matching gradients instead of intensities provides more accurate
reconstructions when illumination is incorrectly estimated under a flatness
prior.
Babak Toghiani-Rizi, Christofer Lind, Maria Svensson, Marcus Windmark
Comments: Results based on a study conducted during the course Intelligent Interactive Systems at Uppsala University, spring 2016
Subjects: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC)
In this report, an automated bartender system was developed for making orders
in a bar using hand gestures. The gesture recognition of the system was
developed using Machine Learning techniques, where the model was trained to
classify gestures using collected data. The final model used in the system
reached an average accuracy of 95%. The system raised ethical concerns both in
terms of user interaction and having such a system in a real world scenario,
but it could initially work as a complement to a real bartender.
Katharina Eggensperger, Marius Lindauer, Frank Hutter
Subjects: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
Good parameter settings are crucial to achieve high performance in many areas
of artificial intelligence (AI), such as satisfiability solving, AI planning,
scheduling, and machine learning (in particular deep learning). Automated
algorithm configuration methods have recently received much attention in the AI
community since they replace tedious, irreproducible and error-prone manual
parameter tuning and can lead to new state-of-the-art performance. However,
practical applications of algorithm configuration are prone to several (often
subtle) pitfalls in the experimental design that can render the procedure
ineffective. We identify several common issues and propose best practices for
avoiding them, including a tool called GenericWrapper4AC for preventing the
many possible problems in measuring the performance of the algorithm being
optimized by executing it in a standardized, controlled manner.
Yanjie Fu, Charu Aggarwal, Srinivasan Parthasarathy, Deepak S. Turaga, Hui Xiong
Comments: To appear in KDD 2017
Subjects: Artificial Intelligence (cs.AI)
Outlier detection is the identification of points in a dataset that do not
conform to the norm. Outlier detection is highly sensitive to the choice of the
detection algorithm and the feature subspace used by the algorithm. Extracting
domain-relevant insights from outliers needs systematic exploration of these
choices since diverse outlier sets could lead to complementary insights. This
challenge is especially acute in an interactive setting, where the choices must
be explored in a time-constrained manner. In this work, we present REMIX, the
first system to address the problem of outlier detection in an interactive
setting. REMIX uses a novel mixed integer programming (MIP) formulation for
automatically selecting and executing a diverse set of outlier detectors within
a time limit. This formulation incorporates multiple aspects such as (i) an
upper limit on the total execution time of detectors (ii) diversity in the
space of algorithms and features, and (iii) meta-learning for evaluating the
cost and utility of detectors. REMIX provides two distinct ways for the analyst
to consume its results: (i) a partitioning of the detectors explored by REMIX
into perspectives through low-rank non-negative matrix factorization; each
perspective can be easily visualized as an intuitive heatmap of experiments
versus outliers, and (ii) an ensembled set of outliers which combines outlier
scores from all detectors. We demonstrate the benefits of REMIX through
extensive empirical validation on real-world data.
Chien-Ping Lu
Comments: 17 pages, 13 figures; to be published in IEEE APSIPA Transaction on Signal and Information Processing as an invited paper on Industrial Technology Advances
Subjects: Artificial Intelligence (cs.AI)
Based on Alan Turing’s proposition on AI and computing machinery, which
shaped Computing as we know it today, I argue that the new AI machine should
understand linear algebra natively. In such a machine, a computing unit does
not need to keep the legacy of a universal computing core. The data can be
delivered to the computing units, and the results can be collected from them
through Collective Streaming, reminiscent of Collective Communication in
Supercomputing. There is no need for a deep memory hierarchy as in a GPU, nor a
fine-grain mesh as in a systolic array.
Ji Young Lee, Franck Dernoncourt, Peter Szolovits
Comments: The first two authors contributed equally to this work
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE); Machine Learning (stat.ML)
Recent approaches based on artificial neural networks (ANNs) have shown
promising results for named-entity recognition (NER). In order to achieve high
performances, ANNs need to be trained on a large labeled dataset. However,
labels might be difficult to obtain for the dataset on which the user wants to
perform NER: label scarcity is particularly pronounced for patient note
de-identification, which is an instance of NER. In this work, we analyze to
what extent transfer learning may address this issue. In particular, we
demonstrate that transferring an ANN model trained on a large labeled dataset
to another dataset with a limited number of labels improves upon the
state-of-the-art results on two different datasets for patient note
de-identification.
Jaeyong Sung, J. Kenneth Salisbury, Ashutosh Saxena
Comments: IEEE International Conference on Robotics and Automation (ICRA), 2017
Subjects: Robotics (cs.RO); Artificial Intelligence (cs.AI); Learning (cs.LG)
The sense of touch, being the earliest sensory system to develop in a human
body [1], plays a critical part of our daily interaction with the environment.
In order to successfully complete a task, many manipulation interactions
require incorporating haptic feedback. However, manually designing a feedback
mechanism can be extremely challenging. In this work, we consider manipulation
tasks that need to incorporate tactile sensor feedback in order to modify a
provided nominal plan. To incorporate partial observation, we present a new
framework that models the task as a partially observable Markov decision
process (POMDP) and learns an appropriate representation of haptic feedback
which can serve as the state for a POMDP model. The model, that is parametrized
by deep recurrent neural networks, utilizes variational Bayes methods to
optimize the approximate posterior. Finally, we build on deep Q-learning to be
able to select the optimal action in each state without access to a simulator.
We test our model on a PR2 robot for multiple tasks of turning a knob until it
clicks.
Ricard Sole
Comments: 3 figures
Subjects: Neurons and Cognition (q-bio.NC); Artificial Intelligence (cs.AI)
The accelerated path of technological development, particularly at the
interface between hardware and biology has been suggested as evidence for
future major technological breakthroughs associated to our potential to
overcome biological constraints. This includes the potential of becoming
immortal, having expanded cognitive capacities thanks to hardware implants or
the creation of intelligent machines. Here I argue that several relevant
evolutionary and structural constraints might prevent achieving most (if not
all) these innovations. Instead, the coming future will bring novelties that
will challenge many other aspects of our life and that can be seen as other
feasible singularities. One particularly important one has to do with the
evolving interactions between humans and non-intelligent robots capable of
learning and communication. Here I argue that a long term interaction can lead
to a new class of “agent” (the humanbot). The way shared memories get tangled
over time will inevitably have important consequences for both sides of the
pair, whose identity as separated entities might become blurred and ultimately
vanish. Understanding such hybrid systems requires a second-order neuroscience
approach while posing serious conceptual challenges, including the definition
of consciousness.
Babak Toghiani-Rizi, Christofer Lind, Maria Svensson, Marcus Windmark
Comments: Results based on a study conducted during the course Intelligent Interactive Systems at Uppsala University, spring 2016
Subjects: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC)
In this report, an automated bartender system was developed for making orders
in a bar using hand gestures. The gesture recognition of the system was
developed using Machine Learning techniques, where the model was trained to
classify gestures using collected data. The final model used in the system
reached an average accuracy of 95%. The system raised ethical concerns both in
terms of user interaction and having such a system in a real world scenario,
but it could initially work as a complement to a real bartender.
Haoyu Xu (1), Chongyang Gu (1 and 2), Han Zhou (2), Junjie Zhang (1) ((1) Chinese Academy of Sciences, Shanghai Advanced Research Institute, China, (2) Shanghai University, School of Communication and Information Engineering, China)
Comments: 13 pages, 4 figures
Subjects: Information Retrieval (cs.IR)
With the rapid development of online recruitment, a large amount of job
postings enables a new paradigm for studying the national economic state.
However, there are two problems when using these data for national economic
research. First, there is a mismatch between the job descriptions and the job
titles. Second, these titles differ from the category names in the national
economic research standards. To map job postings with similar job descriptions
but different titles to the same category of the national economic research
standards, this paper introduces a text classification corpus named HR-CTC. A
novel method to reduce the influence of human subjectivity is proposed to
construct the corpus. The experimental results show that the proposed method
outperforms manual categorization in accuracy evaluation criteria by a 47.08%
increase. To verify the value of this corpus for text classification research
and to provide a baseline for further research, we implement five methods of
deep learning for text classification and achieve promising results.
Darío Garigliotti, Faegheh Hasibi, Krisztian Balog
Comments: Extended version of SIGIR’17 short paper, 5 pages
Subjects: Information Retrieval (cs.IR)
Identifying the target types of entity-bearing queries can help improve
retrieval performance as well as the overall search experience. In this work,
we address the problem of automatically detecting the target types of a query
with respect to a type taxonomy. We propose a supervised learning approach with
a rich variety of features. Using a purpose-built test collection, we show that
our approach outperforms existing methods by a remarkable margin. This is an
extended version of the article published with the same title in the
Proceedings of SIGIR’17.
Ji Young Lee, Franck Dernoncourt, Peter Szolovits
Comments: The first two authors contributed equally to this work
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE); Machine Learning (stat.ML)
Recent approaches based on artificial neural networks (ANNs) have shown
promising results for named-entity recognition (NER). In order to achieve high
performances, ANNs need to be trained on a large labeled dataset. However,
labels might be difficult to obtain for the dataset on which the user wants to
perform NER: label scarcity is particularly pronounced for patient note
de-identification, which is an instance of NER. In this work, we analyze to
what extent transfer learning may address this issue. In particular, we
demonstrate that transferring an ANN model trained on a large labeled dataset
to another dataset with a limited number of labels improves upon the
state-of-the-art results on two different datasets for patient note
de-identification.
Vincent Major, Alisa Surkis, Yindalon Aphinyanaphongs
Subjects: Computation and Language (cs.CL); Machine Learning (stat.ML)
Conventional text classification models make a bag-of-words assumption
reducing text, fundamentally a sequence of words, into word occurrence counts
per document. Recent algorithms such as word2vec and fastText are capable of
learning semantic meaning and similarity between words in an entirely
unsupervised manner using a contextual window and doing so much faster than
previous methods. Each word is represented as a vector such that similar
meaning words such as ‘strong’ and ‘powerful’ are in the same general Euclidian
space. Open questions about these embeddings include their usefulness across
classification tasks and the optimal set of documents to build the embeddings.
In this work, we demonstrate the usefulness of embeddings for improving the
state of the art in classification for our tasks and demonstrate that specific
word embeddings built in the domain and for the tasks can improve performance
over general word embeddings (learnt on news articles, Wikipedia or PubMed).
Katharina Kann, Hinrich Schütze
Subjects: Computation and Language (cs.CL)
We present a semi-supervised way of training a character-based
encoder-decoder recurrent neural network for morphological reinflection, the
task of generating one inflected word form from another. This is achieved by
using unlabeled tokens or random string as training data for an autoencoding
task, adapting a network for morphological reinflection, and performing
multi-task training. We thus use limited labeled data more effectively,
obtaining up to 9.9% improvement over state-of-the-art baselines for 8
different languages.
Wei Wei, Xiaojun Wan
Comments: Accepted by IJCAI 2017
Subjects: Computation and Language (cs.CL); Computers and Society (cs.CY)
Accuracy is one of the basic principles of journalism. However, it is
increasingly hard to manage due to the diversity of news media. Some editors of
online news tend to use catchy headlines which trick readers into clicking.
These headlines are either ambiguous or misleading, degrading the reading
experience of the audience. Thus, identifying inaccurate news headlines is a
task worth studying. Previous work names these headlines “clickbaits” and
mainly focus on the features extracted from the headlines, which limits the
performance since the consistency between headlines and news bodies is
underappreciated. In this paper, we clearly redefine the problem and identify
ambiguous and misleading headlines separately. We utilize class sequential
rules to exploit structure information when detecting ambiguous headlines. For
the identification of misleading headlines, we extract features based on the
congruence between headlines and bodies. To make use of the large unlabeled
data set, we apply a co-training method and gain an increase in performance.
The experiment results show the effectiveness of our methods. Then we use our
classifiers to detect inaccurate headlines crawled from different sources and
conduct a data analysis.
Xu Tian, Jun Zhang, Zejun Ma, Yi He, Juan Wei
Comments: 5 pages
Subjects: Computation and Language (cs.CL)
Frame stacking is broadly applied in end-to-end neural network training like
connectionist temporal classification (CTC), and it leads to more accurate
models and faster decoding. However, it is not well-suited to conventional
neural network based on context-dependent state acoustic model, if the decoder
is unchanged. In this paper, we propose a novel frame retaining method which is
applied in decoding. The system which combined frame retaining with frame
stacking could reduces the time consumption of both training and decoding. Long
short-term memory (LSTM) recurrent neural networks (RNNs) using it achieve
almost linear training speedup and reduces relative 41\% real time factor
(RTF). At the same time, recognition performance is no degradation or improves
sightly on Shenma voice search dataset in Mandarin.
Dat Quoc Nguyen, Mark Dras, Mark Johnson
Subjects: Computation and Language (cs.CL)
We present a novel neural network model that learns POS tagging and
graph-based dependency parsing jointly. Our model uses bidirectional LSTMs to
learn feature representations shared for both POS tagging and dependency
parsing tasks, thus handling the feature-engineering problem. Our extensive
experiments, on 19 languages from the Universal Dependencies project, show that
our model outperforms the state-of-the-art neural network-based
Stack-propagation model for joint POS tagging and transition-based dependency
parsing, resulting in a new state of the art. Our code is open-source and
available at: this https URL
Enes Avcu, Chihiro Shibata, Jeffrey Heinz
Subjects: Computation and Language (cs.CL)
This paper presents experiments illustrating how formal language theory can
shed light on deep learning. We train naive Long Short-Term Memory (LSTM)
Recurrent Neural Networks (RNNs) on six formal languages drawn from the
Strictly Local (SL) and Strictly Piecewise (SP) classes. These classes are
relevant to computational linguistics and among the simplest in a
mathematically well-understood hierarchy of subregular classes. SL and SP
classes encode local and long-distance dependencies, respectively. The results
show four of the six languages were learned remarkably well, but overfitting
arguably occurred with the simplest SL language and undergeneralization with
the most complex SP pattern. Even though LSTMs were developed to handle
long-distance dependencies, the latter result shows they stymie naive LSTMs in
contrast to local dependencies. While it remains to be seen which of the many
variants of LSTMs may learn SP languages well, this result speaks to the larger
point that the judicial use of formal language theory can illuminate the inner
workings of RNNs.
Tristan Konolige, Jed Brown
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC)
Problems from graph drawing, spectral clustering, network flow and graph
parti- tioning all can be expressed as Laplacian matrices. Theoretically fast
approaches to solving these problems exist, but in practice these techniques
are slow. Three practical approaches have been proposed and work well in
serial. However, as problem sizes increase and single core speeds stagnate,
parallelism is essential to solve problems quickly. We present an unsmoothed
aggregation Multigrid method for solving graph Laplacians in distributed memory
setting. Our solver scales up to 64 compute nodes and achieves speedups of up
to 83x over the existing serial solutions.
Derek Weitzel, Brian Bockelman, Duncan A. Brown, Peter Couvares, Frank Würthwein, Edgar Fajardo Hernandez
Comments: 6 pages, 3 figures, submitted to PEARC17
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC); Instrumentation and Methods for Astrophysics (astro-ph.IM)
During 2015 and 2016, the Laser Interferometer Gravitational-Wave Observatory
(LIGO) conducted a three-month observing campaign. These observations delivered
the first direct detection of gravitational waves from binary black hole
mergers. To search for these signals, the LIGO Scientific Collaboration uses
the PyCBC search pipeline. To deliver science results in a timely manner, LIGO
collaborated with the Open Science Grid (OSG) to distribute the required
computation across a series of dedicated, opportunistic, and allocated
resources. To deliver the petabytes necessary for such a large-scale
computation, our team deployed a distributed data access infrastructure based
on the XRootD server suite and the CernVM File System (CVMFS). This data access
strategy grew from simply accessing remote storage to a POSIX-based interface
underpinned by distributed, secure caches across the OSG.
Christoph Lenzen, Joel Rybicki
Comments: 56+2 pages, 12 figures
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC)
We give fault-tolerant algorithms for establishing synchrony in distributed
systems in which each of the (n) nodes has its own clock. Our algorithms
operate in a very strong fault model: we require self-stabilisation, i.e., the
initial state of the system may be arbitrary, and there can be up to (f<n/3)
ongoing Byzantine faults, i.e., nodes that deviate from the protocol in an
arbitrary manner. Furthermore, we assume that the local clocks of the nodes may
progress at different speeds (clock drift) and communication has bounded delay.
In this model, we study the pulse synchronisation problem, where the task is to
guarantee that eventually all correct nodes generate well-separated local pulse
events (i.e., unlabelled logical clock ticks) in a synchronised manner.
Compared to prior work, we achieve exponential improvements in stabilisation
time and the number of communicated bits, and give the first sublinear-time
algorithm for the problem:
– In the deterministic setting, the state-of-the-art solutions stabilise in
time (Theta(f)) and have each node broadcast (Theta(f log f)) bits per time
unit. We exponentially reduce the number of bits broadcasted per time unit to
(Theta(log f)) while retaining the same stabilisation time.
– In the randomised setting, the state-of-the-art solutions stabilise in time
(Theta(f)) and have each node broadcast (O(1)) bits per time unit. We
exponentially reduce the stabilisation time to (log^{O(1)} f) while each node
broadcasts (log^{O(1)} f) bits per time unit.
These results are obtained by means of a recursive approach reducing the
above task of self-stabilising pulse synchronisation in the bounded-delay model
to non-self-stabilising binary consensus in the synchronous model. In general,
our approach introduces at most logarithmic overheads in terms of stabilisation
time and broadcasted bits over the underlying consensus routine.
Chen Avin, Kaushik Mondal, Stefan Schmid
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC)
Traditionally, networks such as datacenter interconnects are designed to
optimize worst-case performance under arbitrary traffic patterns. Such network
designs can however be far from optimal when considering the actual workloads
and traffic patterns which they serve. This insight led to the development of
demand-aware datacenter interconnects which can be reconfigured depending on
the workload.
Motivated by these trends, this paper initiates the algorithmic study of
demand-aware networks (DANs) designs, and in particular the design of
bounded-degree networks. The inputs to the network design problem are a
discrete communication request distribution, D, defined over communicating
pairs from the node set V , and a bound, d, on the maximum degree. In turn, our
objective is to design an (undirected) demand-aware network N = (V,E) of
bounded-degree d, which provides short routing paths between frequently
communicating nodes distributed across N. In particular, the designed network
should minimize the expected path length on N (with respect to D), which is a
basic measure of the efficiency of the network.
We show that this fundamental network design problem exhibits interesting
connections to several classic combinatorial problems and to information
theory. We derive a general lower bound based on the entropy of the
communication pattern D, and present asymptotically optimal network-aware
design algorithms for important distribution families, such as sparse
distributions and distributions of locally bounded doubling dimensions.
Thomas D. Dickerson
Comments: pre-print, submitted to DISC’17
Subjects: Data Structures and Algorithms (cs.DS); Distributed, Parallel, and Cluster Computing (cs.DC)
This paper proposes a new concurrent heap algorithm, based on a stateless
shape property, which efficiently maintains balance during insert and removeMin
operations implemented with hand-over-hand locking. It also provides a O(1)
linearizable snapshot operation based on lazy copy-on-write semantics. Such
snapshots can be used to provide consistent views of the heap during iteration,
as well as to make speculative updates (which can later be dropped).
The simplicity of the algorithm allows it to be easily proven correct, and
the choice of shape property provides priority queue performance which is
competitive with highly optimized skiplist implementations (and has stronger
bounds on worst-case time complexity).
A Scala reference implementation is provided.
Gautam Bhanage
Comments: 6 pages, 7 figures
Subjects: Networking and Internet Architecture (cs.NI); Distributed, Parallel, and Cluster Computing (cs.DC)
This study presents the design of the hybrid wireless virtualization HWV
controller based network architecture. Using a HWV controller, an unified
approach can be taken for provisioning and management of virtualized
heterogeneous radios, irrespective of their MAC and PHY layer mechanisms. It is
shown that the airtime occupancy by transmissions from different slices or
groups can be used as a single metric for tying these virtualized platforms.
The HWV controller can account and dynamically reprovision slice quotas, which
can be used for maximizing the revenue of the network operator or aggregate
system throughput performance. Results from simulations show that an HWV
controller based infrastructure is able to improve the revenue generated from a
single virtualized basestation and an AP by up to 40 percent under tested
conditions.
Kleomenis Katevas, Ilias Leontiadis, Martin Pielot, Joan Serrà
Comments: 6 pages, 3 figures, 3 tables
Subjects: Learning (cs.LG)
We present a practical approach for processing mobile sensor time series data
for continual deep learning predictions. The approach comprises data cleaning,
normalization, capping, time-based compression, and finally classification with
a recurrent neural network. We demonstrate the effectiveness of the approach in
a case study with 279 participants. On the basis of sparse sensor events, the
network continually predicts whether the participants would attend to a
notification within 10 minutes. Compared to a random baseline, the classifier
achieves a 40% performance increase (AUC of 0.702) on a withheld test set. This
approach allows to forgo resource-intensive, domain-specific, error-prone
feature engineering, which may drastically increase the applicability of
machine learning to mobile phone sensor data.
Jonas Moritz Kohler, Aurelien Lucchi
Comments: ICML 2017
Subjects: Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
We consider the minimization of non-convex functions that typically arise in
machine learning. Specifically, we focus our attention on a variant of trust
region methods known as cubic regularization. This approach is particularly
attractive because it escapes strict saddle points and it provides stronger
convergence guarantees than first- and second-order as well as classical trust
region methods. However, it suffers from a high computational complexity that
makes it impractical for large-scale learning. Here, we propose a novel method
that uses sub-sampling to lower this computational cost. By the use of
concentration inequalities we provide a sampling scheme that gives sufficiently
accurate gradient and Hessian approximations to retain the strong global and
local convergence guarantees of cubically regularized methods. To the best of
our knowledge this is the first work that gives global convergence guarantees
for a sub-sampled variant of cubic regularization on non-convex functions.
Furthermore, we provide experimental results supporting our theory.
Jaeyong Sung, J. Kenneth Salisbury, Ashutosh Saxena
Comments: IEEE International Conference on Robotics and Automation (ICRA), 2017
Subjects: Robotics (cs.RO); Artificial Intelligence (cs.AI); Learning (cs.LG)
The sense of touch, being the earliest sensory system to develop in a human
body [1], plays a critical part of our daily interaction with the environment.
In order to successfully complete a task, many manipulation interactions
require incorporating haptic feedback. However, manually designing a feedback
mechanism can be extremely challenging. In this work, we consider manipulation
tasks that need to incorporate tactile sensor feedback in order to modify a
provided nominal plan. To incorporate partial observation, we present a new
framework that models the task as a partially observable Markov decision
process (POMDP) and learns an appropriate representation of haptic feedback
which can serve as the state for a POMDP model. The model, that is parametrized
by deep recurrent neural networks, utilizes variational Bayes methods to
optimize the approximate posterior. Finally, we build on deep Q-learning to be
able to select the optimal action in each state without access to a simulator.
We test our model on a PR2 robot for multiple tasks of turning a knob until it
clicks.
Albert S. Berahas, Raghu Bollapragada, Jorge Nocedal
Comments: 25 pages, 22 figures
Subjects: Optimization and Control (math.OC); Learning (cs.LG); Machine Learning (stat.ML)
The concepts of sketching and subsampling have recently received much
attention by the optimization and statistics communities. In this paper, we
study Newton-Sketch and Subsampled Newton (SSN) methods for the finite-sum
optimization problem. We consider practical versions of the two methods in
which the Newton equations are solved approximately using the conjugate
gradient (CG) method or a stochastic gradient iteration. We establish new
complexity results for the SSN-CG method that exploit the spectral properties
of CG. Controlled numerical experiments compare the relative strengths of
Newton-Sketch and SSN methods and show that for many finite-sum problems, they
are far more efficient than SVRG, a popular first-order method.
Abhishek Sharma
Comments: 8 pages, Under Review
Subjects: Computer Vision and Pattern Recognition (cs.CV); Learning (cs.LG)
This paper presents a novel framework in which image cosegmentation and
colocalization are cast into a single optimization problem that integrates
information from low level appearance cues with that of high level localization
cues in a very weakly supervised manner. In contrast to multi-task learning
paradigm that learns similar tasks using a shared representation, the proposed
framework leverages two representations at different levels and simultaneously
discriminates between foreground and background at the bounding box and
superpixel level using discriminative clustering. We show empirically that
constraining the two problems at different scales enables the transfer of
semantic localization cues to improve cosegmentation output whereas local
appearance based segmentation cues help colocalization. The unified framework
outperforms strong baseline approaches, of learning the two problems
separately, by a large margin on four benchmark datasets. Furthermore, it
obtains competitive results compared to the state of the art for cosegmentation
on two benchmark datasets and second best result for colocalization on Pascal
VOC 2007.
Maxat Kulmanov, Mohammed Asif Khan, Robert Hoehndorf
Subjects: Genomics (q-bio.GN); Learning (cs.LG); Quantitative Methods (q-bio.QM)
A large number of protein sequences are becoming available through the
application of novel high-throughput sequencing technologies. Experimental
functional characterization of these proteins is time-consuming and expensive,
and is often only done rigorously for few selected model organisms.
Computational function prediction approaches have been suggested to fill this
gap. The functions of proteins are classified using the Gene Ontology (GO),
which contains over 40,000 classes. Additionally, proteins have multiple
functions, making function prediction a large-scale, multi-class, multi-label
problem.
We have developed a novel method to predict protein function from sequence.
We use deep learning to learn features from protein sequences as well as a
cross-species protein-protein interaction network. Our approach specifically
outputs information in the structure of the GO and utilizes the dependencies
between GO classes as background information to construct a deep learning
model. We evaluate our method using the standards established by the
Computational Assessment of Function Annotation (CAFA) and demonstrate a
significant improvement over baseline methods such as BLAST, with significant
improvement for predicting cellular locations.
Nikolai Dokuchaev
Subjects: Information Theory (cs.IT)
The paper studies functions defined on continuous branching lines connected
into a system. A notion of spectrum degeneracy for these systems of functions
is introduced. This degeneracy is based on the properties of the Fourier
transforms for processes representing functions on the branches that are deemed
to be extended onto the real axis. This spectrum degeneracy ensures some
opportunities for extrapolation and sampling. The topology of the system is
taken into account via a restriction that these processes coincides on certain
parts of real axis. It is shown that processes with this feature are everywhere
dense in the set of processes equivalent to functions on the branching lines.
Some applications to extrapolation and sampling are considered.
Zheng Chu, Tuan Anh Le, Huan X. Nguyen, Arumugam Nallanathan, Mehmet Karamanoglu
Comments: 5 pages, 3 figures, to appear in Proc. VTC Fall, Toronto, Canada, Sept. 2017
Subjects: Information Theory (cs.IT)
This paper considers multiple-input multiple-output (MIMO) full-duplex (FD)
two-way secrecy systems. Specifically, both multi-antenna FD legitimate nodes
exchange their own confidential message in the presence of an eavesdropper.
Taking into account the imperfect channel state information (CSI) of the
eavesdropper, we formulate a robust sum secrecy rate maximization (RSSRM)
problem subject to the outage probability constraint of the achievable sum
secrecy rate and the transmit power constraint. Unlike other existing channel
uncertainty models, e.g., norm-bounded and Gaussian-distribution, we exploit a
moment-based random distributed CSI channel uncertainty model to recast our
formulate RSSRM problem into the convex optimization frameworks based on a
Markov’s inequality and robust conic reformulation, i.e., semidefinite
programming (SDP). In addition, difference-of-concave (DC) approximation is
employed to iteratively tackle the transmit covariance matrices of these
legitimate nodes. Simulation results are provided to validate our proposed FD
approaches.
Thomas Lundgaard Hansen, Bernard Henri Fleury, Bhaskar D. Rao
Comments: 13 pages, 4 figures, submitted to IEEE Transactions on Signal Processing
Subjects: Information Theory (cs.IT); Applications (stat.AP)
A number of recent works have proposed to solve the line spectral estimation
problem by applying an off-the-grid ex- tension of sparse estimation
techniques. These methods are more advantageous than classical line spectral
estimation algorithms because they inherently estimate the model order.
However, they all have computation times which grow at least cubically in the
problem size, which limits their practical applicability for large problem
sizes. To alleviate this issue, we propose a low-complexity method for line
spectral estimation, which also draws on ideas from sparse estimation. Our
method is based on a probabilistic view of the problem. The signal covariance
matrix is shown to have Toeplitz structure, allowing superfast Toeplitz
inversion to be used. We demonstrate that our method achieves estimation
accuracy at least as good as current methods and that it does so while being
orders of magnitudes faster.
Ming Xiao, Shahid Mumtaz, Yongming Huang, Linglong Dai, Yonghui Li, Michail Matthaiou, George K. Karagiannidis, Emil Björnson, Kai Yang, Chih Lin, Amitava Ghosh
Subjects: Information Theory (cs.IT)
Millimeter wave (mmWave) communications have recently attracted large
research interest, since the huge available bandwidth can potentially lead to
rates of multiple Gbps (gigabit per second) per user. Though mmWave can be
readily used in stationary scenarios such as indoor hotspots or backhaul, it is
challenging to use mmWave in mobile networks, where the transmitting/receiving
nodes may be moving, channels may have a complicated structure, and the
coordination among multiple nodes is difficult. To fully exploit the high
potential rates of mmWave in mobile networks, lots of technical problems must
be addressed. This paper presents a comprehensive survey of mmWave
communications for future mobile networks (5G and beyond). We first summarize
the recent channel measurement campaigns and modeling results. Then, we discuss
in detail recent progresses in multiple input multiple output (MIMO)
transceiver design for mmWave communications. After that, we provide an
overview of the solution for multiple access and backhauling, followed by
analysis of coverage and connectivity. Finally, the progresses in the
standardization and deployment of mmWave for mobile networks are discussed.
Irwansyah, Intan Muchtadi-Alamsyah, Ahmad Muchlis, Aleams Barra, Djoko Suprijanto
Subjects: Information Theory (cs.IT)
In this paper we give the enumeration formulas for Euclidean self-dual
skew-cyclic codes over finite fields when ((n,| heta|)=1) and for some cases
when ((n,| heta|)>1,) where (n) is the length of the code and (| heta|) is
the order of automorphism ( heta.)
Aiden Price, Joanne Hall
Subjects: Information Theory (cs.IT)
LDPC codes are used in many applications, however, their error correcting
capabilities are limited by the presence of stopping sets and trapping sets.
Trapping sets and stopping sets occur when specific low-wiehgt error patterns
cause a decoder to fail. Trapping sets were first discovered with investigation
of the error floor of the Margulis code. Possible solutions are constructions
which avoid creating trapping sets, such as progressive edge growth (PEG), or
methods which remove trapping sets from existing constructions, such as graph
covers. This survey examines trapping sets and stopping sets in LDPC codes over
channels such as BSC, BEC and AWGNC.
Jincheng Dai, Kai Niu, Zhongwei Si, Chao Dong, Jiaru Lin
Comments: First version
Subjects: Information Theory (cs.IT)
Non-orthogonal multiple access (NOMA) is one of the key techniques to address
the high spectral efficiency and massive connectivity requirements for the
fifth generation (5G) wireless system. To efficiently realize NOMA, we propose
a joint design framework combining the polar coding and the NOMA transmission,
which deeply mines the generalized polarization effect among the users. In this
polar coded NOMA (PC-NOMA) framework, the original NOMA channel is decomposed
into multiple bit polarized channels by using a three-stage channel transform,
that is, user( o)signal( o)bit partitions. Specifically, for the first-stage
channel transform, we design two schemes, namely sequential user partition
(SUP) and parallel user partition (PUP). For the SUP, a joint successive
cancellation detecting and decoding scheme is developed, and a search algorithm
is proposed to schedule the NOMA detecting order which improves the system
performance by enhanced polarization among the user synthesized channels. The
“worst-goes-first” idea is employed in the scheduling strategy, and its
theoretic performance is analyzed by using the polarization principle. For the
PUP, a corresponding parallel detecting scheme is exploited to reduce the
latency. The block error ratio performances over the additive white Gaussian
noise channel and the Rayleigh fading channel indicate that the proposed
PC-NOMA obviously outperforms the state-of-the-art turbo coded NOMA scheme due
to the advantages of joint design between the polar coding and NOMA.
Zhiqiang Wei, Derrick Wing Kwan Ng, Jinhong Yuan, Hui-Ming Wang
Comments: Accepted for publication, IEEE TCOM, May 17, 2017
Subjects: Information Theory (cs.IT)
In this paper, we study power-efficient resource allocation for multicarrier
non-orthogonal multiple access (MC-NOMA) systems. The resource allocation
algorithm design is formulated as a non-convex optimization problem which
jointly designs the power allocation, rate allocation, user scheduling, and
successive interference cancellation (SIC) decoding policy for minimizing the
total transmit power. The proposed framework takes into account the
imperfection of channel state information at transmitter (CSIT) and quality of
service (QoS) requirements of users. To facilitate the design of optimal SIC
decoding policy on each subcarrier, we define a channel-to-noise ratio outage
threshold. Subsequently, the considered non-convex optimization problem is
recast as a generalized linear multiplicative programming problem, for which a
globally optimal solution is obtained via employing the branch-and-bound
approach. The optimal resource allocation policy serves as a system performance
benchmark due to its high computational complexity. To strike a balance between
system performance and computational complexity, we propose a suboptimal
iterative resource allocation algorithm based on difference of convex
programming. Simulation results demonstrate that the suboptimal scheme achieves
a close-to-optimal performance. Also, both proposed schemes provide significant
transmit power savings than that of conventional orthogonal multiple access
(OMA) schemes.
Vutha Va, Junil Choi, Takayuki Shimizu, Gaurav Bansal, Robert W. Heath Jr
Comments: 14 pages, 17 figures, Submitted to IEEE Transactions on Vehicular Technology
Subjects: Information Theory (cs.IT)
Efficient beam alignment is a crucial component in millimeter wave systems
with analog beamforming, especially in fast-changing vehicular settings. This
paper uses the vehicle’s position (e.g., available via GPS) to query the
multipath fingerprint database, which provides prior knowledge of potential
pointing directions for reliable beam alignment. The approach is the inverse of
fingerprinting localization, where the measured multipath signature is compared
to the fingerprint database to retrieve the most likely position. The power
loss probability is introduced as a metric to quantify misalignment accuracy
and is used for optimizing candidate beam selection. Two candidate beam
selection methods are derived, where one is a heuristic while the other
minimizes the misalignment probability. The proposed beam alignment is
evaluated using realistic channels generated from a commercial ray-tracing
simulator. Using the generated channels, an extensive investigation is
provided, which includes the required measurement sample size to build an
effective fingerprint, the impact of measurement noise, the sensitivity to
changes in traffic density, and a beam alignment overhead comparison with IEEE
802.11ad as the baseline. Using the concept of beam coherence time, which is
the duration between two consecutive beam alignments, and parameters of IEEE
802.11ad, the overhead is compared in the mobility context. The results show
that while the proposed approach provides increasing rates with larger antenna
arrays, IEEE 802.11ad has decreasing rates due to the larger beam training
overhead that eats up a large portion of the beam coherence time, which becomes
shorter with increasing mobility.
Masahito Hayashi
Subjects: Statistics Theory (math.ST); Information Theory (cs.IT)
We consider the estimation of hidden Markovian process by using information
geometry with respect to transition matrices. We consider the case when we use
only the histogram of (k)-memory data. Firstly, we focus on a partial
observation model with Markovian process and we show that the asymptotic
estimation error of this model is given as the inverse of projective Fisher
information of transition matrices. Next, we apply this result to the
estimation of hidden Markovian process. For this purpose, we define an
exponential family of ({cal Y})-valued transition matrices. We carefully
discuss the equivalence problem for hidden Markovian process on the tangent
space. Then, we propose a novel method to estimate hidden Markovian process.