scholarly journals Session Recommendation via Recurrent Neural Networks over Fisher Embedding Vectors

Sensors ◽  
2019 ◽  
Vol 19 (16) ◽  
pp. 3498
Author(s):  
Domokos Kelen ◽  
Bálint Daróczy ◽  
Frederick Ayala-Gómez ◽  
Anna Ország ◽  
András Benczúr
Keyword(s):  
Web Sites ◽  
Performance Metrics ◽  
Transition Probabilities ◽  
Feature Representation ◽  
Distance Measures ◽  
Data Sets ◽  
Positioning Systems ◽  
New Model ◽  
Location Based Social Networks ◽  
Recent Item

Recommendation services bear great importance in e-commerce, shopping, tourism, and social media, as they aid the user in navigating through the items that are most relevant to their needs. In order to build recommender systems, organizations log the item consumption in their user sessions by using different sensors. For instance, Web sites use Web data loggers, museums and shopping centers rely on user in-door positioning systems to register user movement, and Location-Based Social Networks use Global Positioning System for out-door user tracking. Most organizations do not have a detailed history of previous activities or purchases by the user. Hence, in most cases recommenders propose items that are similar to the most recent ones viewed in the current user session. The corresponding task is called session based, and when only the last item is considered, it is referred to as item-to-item recommendation. A natural way of building next-item recommendations relies on item-to-item similarities and item-to-item transitions in the form of “people who viewed this, also viewed” lists. Such methods, however, depend on local information for the given item pairs, which can result in unstable results for items with short transaction history, especially in connection with the cold-start items that recently appeared and had no time yet to accumulate a sufficient number of transactions. In this paper, we give new algorithms by defining a global probabilistic similarity model of all the items based on Random Fields. We give a generative model for the item interactions based on arbitrary distance measures over the items, including explicit, implicit ratings and external metadata to estimate and predict item-to-item transition probabilities. We exploit our new model in two different item similarity algorithms, as well as a feature representation in a recurrent neural network based recommender. Our experiments on various publicly available data sets show that our new model outperforms simple similarity baseline methods and combines well with recent item-to-item and deep learning recommenders under several different performance metrics.

scholarly journals An edge-cloud collaboration architecture for pattern anomaly detection of time series in wireless sensor networks

2021 ◽  
Author(s):  
Cong Gao ◽  
Ping Yang ◽  
Yanping Chen ◽  
Zhongmin Wang ◽  
Yue Wang
Keyword(s):  
Time Series ◽  
Wireless Sensor Networks ◽  
Sensor Networks ◽  
Anomaly Detection ◽  
Estimation Method ◽  
Feature Representation ◽  
Sensor Data ◽  
Wireless Sensor ◽  
Data Sets ◽  
Edge Node

AbstractWith large deployment of wireless sensor networks, anomaly detection for sensor data is becoming increasingly important in various fields. As a vital data form of sensor data, time series has three main types of anomaly: point anomaly, pattern anomaly, and sequence anomaly. In production environments, the analysis of pattern anomaly is the most rewarding one. However, the traditional processing model cloud computing is crippled in front of large amount of widely distributed data. This paper presents an edge-cloud collaboration architecture for pattern anomaly detection of time series. A task migration algorithm is developed to alleviate the problem of backlogged detection tasks at edge node. Besides, the detection tasks related to long-term correlation and short-term correlation in time series are allocated to cloud and edge node, respectively. A multi-dimensional feature representation scheme is devised to conduct efficient dimension reduction. Two key components of the feature representation trend identification and feature point extraction are elaborated. Based on the result of feature representation, pattern anomaly detection is performed with an improved kernel density estimation method. Finally, extensive experiments are conducted with synthetic data sets and real-world data sets.

scholarly journals A Mine of Information: Can Sports Analytics Provide Wisdom From Your Data?

2017 ◽  
Vol 12 (7) ◽  
pp. 851-855 ◽  
Author(s):  
Louis Passfield ◽  
James G. Hopker
Keyword(s):  
Web Sites ◽  
Large Data ◽  
Objective Evaluation ◽  
Large Data Sets ◽  
Data Sets ◽  
Competitive Strategies ◽  
Data Repositories ◽  
Sports Analytics ◽  
Sport And Exercise ◽  
New Research

This paper explores the notion that the availability and analysis of large data sets have the capacity to improve practice and change the nature of science in the sport and exercise setting. The increasing use of data and information technology in sport is giving rise to this change. Web sites hold large data repositories, and the development of wearable technology, mobile phone applications, and related instruments for monitoring physical activity, training, and competition provide large data sets of extensive and detailed measurements. Innovative approaches conceived to more fully exploit these large data sets could provide a basis for more objective evaluation of coaching strategies and new approaches to how science is conducted. An emerging discipline, sports analytics, could help overcome some of the challenges involved in obtaining knowledge and wisdom from these large data sets. Examples of where large data sets have been analyzed, to evaluate the career development of elite cyclists and to characterize and optimize the training load of well-trained runners, are discussed. Careful verification of large data sets is time consuming and imperative before useful conclusions can be drawn. Consequently, it is recommended that prospective studies be preferred over retrospective analyses of data. It is concluded that rigorous analysis of large data sets could enhance our knowledge in the sport and exercise sciences, inform competitive strategies, and allow innovative new research and findings.

scholarly journals Interactive coupling of regional atmosphere with biosphere in the new generation regional climate system model REMO-iMOVE

2014 ◽  
Vol 7 (3) ◽  
pp. 1093-1114 ◽  
Author(s):  
C. Wilhelm ◽  
D. Rechid ◽  
D. Jacob
Keyword(s):  
Land Surface ◽  
Regional Climate ◽  
Regional Model ◽  
Horizontal Resolution ◽  
Soil Conditions ◽  
Data Sets ◽  
Vegetation Phenology ◽  
Land Surface Scheme ◽  
New Model ◽  
Surface Scheme

Abstract. The main objective of this study is the coupling of the regional climate model REMO with a new land surface scheme including dynamic vegetation phenology, and the evaluation of the new model version called REMO with interactive MOsaic-based VEgetation: REMO-iMOVE. First, we focus on the documentation of the technical aspects of the new model constituents and the coupling mechanism. The representation of vegetation in iMOVE is based on plant functional types (PFTs). Their geographical distribution is prescribed to the model which can be derived from different land surface data sets. Here, the PFT distribution is derived from the GLOBCOVER 2000 data set which is available on 1 km × 1 km horizontal resolution. Plant physiological processes like photosynthesis, respiration and transpiration are incorporated into the model. The vegetation modules are fully coupled to atmosphere and soil. In this way, plant physiological activity is directly driven by atmospheric and soil conditions at the model time step (two minutes to some seconds). In turn, the vegetation processes and properties influence the exchange of substances, energy and momentum between land and atmosphere. With the new coupled regional model system, dynamic feedbacks between vegetation, soil and atmosphere are represented at regional to local scale. In the evaluation part, we compare simulation results of REMO-iMOVE and of the reference version REMO2009 to multiple observation data sets of temperature, precipitation, latent heat flux, leaf area index and net primary production, in order to investigate the sensitivity of the regional model to the new land surface scheme and to evaluate the performance of both model versions. Simulations for the regional model domain Europe on a horizontal resolution of 0.44° had been carried out for the time period 1995–2005, forced with ECMWF ERA-Interim reanalyses data as lateral boundary conditions. REMO-iMOVE is able to simulate the European climate with the same quality as the parent model REMO2009. Differences in near-surface climate parameters can be restricted to some regions and are mainly related to the new representation of vegetation phenology. The seasonal and interannual variations in growth and senescence of vegetation are captured by the model. The net primary productivity lies in the range of observed values for most European regions. This study reveals the need for implementing vertical soil water dynamics in order to differentiate the access of plants to water due to different rooting depths. This gets especially important if the model will be used in dynamic vegetation studies.

scholarly journals AA-HMM: An Anti-Adversarial Hidden Markov Model for Network-Based Intrusion Detection

2018 ◽  
Vol 8 (12) ◽  
pp. 2421 ◽  
Author(s):  
Chongya Song ◽  
Alexander Pons ◽  
Kang Yen
Keyword(s):  
Markov Model ◽  
Hidden Markov Model ◽  
Performance Metrics ◽  
Hidden Markov ◽  
Data Sets ◽  
Learning Abilities ◽  
Benchmark Data ◽  
Malicious Behavior ◽  
Network Intrusion ◽  
The Common

In the field of network intrusion, malware usually evades anomaly detection by disguising malicious behavior as legitimate access. Therefore, detecting these attacks from network traffic has become a challenge in this an adversarial setting. In this paper, an enhanced Hidden Markov Model, called the Anti-Adversarial Hidden Markov Model (AA-HMM), is proposed to effectively detect evasion pattern, using the Dynamic Window and Threshold techniques to achieve adaptive, anti-adversarial, and online-learning abilities. In addition, a concept called Pattern Entropy is defined and acts as the foundation of AA-HMM. We evaluate the effectiveness of our approach employing two well-known benchmark data sets, NSL-KDD and CTU-13, in terms of the common performance metrics and the algorithm’s adaptation and anti-adversary abilities.

Mixture copulas and insurance applications

2018 ◽  
Vol 12 (2) ◽  
pp. 391-411
Author(s):  
Maissa Tamraz
Keyword(s):  
Joint Distribution ◽  
Stable Distribution ◽  
Random Variable ◽  
Fixed Time ◽  
Collective Model ◽  
Data Sets ◽  
New Model ◽  
Time Period ◽  
Insurance Data ◽  
Dependence Property

AbstractIn the classical collective model over a fixed time period of two insurance portfolios, we are interested, in this contribution, in the models that relate to the joint distributionFof the largest claim amounts observed in both insurance portfolios. Specifically, we consider the tractable model where the claim counting random variableNfollows a discrete-stable distribution with parameters (α,λ). We investigate the dependence property ofFwith respect to both parametersαandλ. Furthermore, we present several applications of the new model to concrete insurance data sets and assess the fit of our new model with respect to other models already considered in some recent contributions. We can see that our model performs well with respect to most data sets.

scholarly journals Collecting public RGB-D datasets for human daily activity recognition

2017 ◽  
Vol 14 (4) ◽  
pp. 172988141770907 ◽  
Author(s):  
Hanbo Wu ◽  
Xin Ma ◽  
Zhimeng Zhang ◽  
Haibo Wang ◽  
Yibin Li
Keyword(s):  
Activity Recognition ◽  
Daily Activity ◽  
Visual Cues ◽  
Large Scale ◽  
Hot Spot ◽  
Feature Representation ◽  
Data Sets ◽  
Activity Data ◽  
Data Set ◽  
Depth Motion Maps

Human daily activity recognition has been a hot spot in the field of computer vision for many decades. Despite best efforts, activity recognition in naturally uncontrolled settings remains a challenging problem. Recently, by being able to perceive depth and visual cues simultaneously, RGB-D cameras greatly boost the performance of activity recognition. However, due to some practical difficulties, the publicly available RGB-D data sets are not sufficiently large for benchmarking when considering the diversity of their activities, subjects, and background. This severely affects the applicability of complicated learning-based recognition approaches. To address the issue, this article provides a large-scale RGB-D activity data set by merging five public RGB-D data sets that differ from each other on many aspects such as length of actions, nationality of subjects, or camera angles. This data set comprises 4528 samples depicting 7 action categories (up to 46 subcategories) performed by 74 subjects. To verify the challengeness of the data set, three feature representation methods are evaluated, which are depth motion maps, spatiotemporal depth cuboid similarity feature, and curvature space scale. Results show that the merged large-scale data set is more realistic and challenging and therefore more suitable for benchmarking.

Do Common User Interface Design Patterns Improve Navigation?

2002 ◽  
Vol 46 (14) ◽  
pp. 1315-1319 ◽  
Author(s):  
Carlos A. Maldonado ◽  
Marc L. Jlesnick
Keyword(s):  
Web Sites ◽  
Interface Design ◽  
Design Patterns ◽  
Performance Metrics ◽  
Statistical Significance ◽  
A Priori ◽  
Improve Performance ◽  
Subjective Preference ◽  
On Line ◽  
Quantitative Performance

The Internet has become a growing channel for consumer purchases. Half of all U.S. consumers made at least one purchase on-line in 2001. However, many consumers report frustration with the lack of support for navigation within many Internet retailers' web sites. Several design patterns have been suggested to overcome these limitations, such as expanded hierarchies and breadcrumbs. This study investigated the effects of these design patterns on users' quantitative performance and subjective preference for ecommerce web sites. Expanded hierarchies, a design pattern that is commonly used by many retail web sites, degraded all of the performance metrics assessed in the study. Users required more time, made more errors, used more clicks, and had lower satisfaction scores for sites designed with expanded hierarchies. The results for breadcrumbs suggest that they may improve performance. The inclusion of breadcrumbs reduced the number of clicks required by users to complete the tasks, but other performance metrics did not reach statistical significance. The results indicate that design patterns that are believed to improve performance a priori may not yield the results expected.

Approximate Matching Between XML Documents and Schemas with Applications in XML Classification and Clustering

2011 ◽  
pp. 99-124 ◽  
Author(s):  
Guangming Xing
Keyword(s):  
Structural Information ◽  
Experimental Studies ◽  
Feature Representation ◽  
Document Management ◽  
Data Sets ◽  
Structural Classification ◽  
Approximate Matching ◽  
Xml Documents ◽  
Efficiency And Effectiveness ◽  
The Cost

Classification/clustering of XML documents based on their structural information is important for many tasks related with document management. In this chapter, we present a suite of algorithms to compute the cost for approximate matching between XML documents and schemas. A framework for classifying/clustering XML documents by structure is then presented based on the computation of distances between XML documents and schemas. The backbone of the framework is the feature representation using a vector of the distances. Experimental studies were conducted on various XML data sets, suggesting the efficiency and effectiveness of our approach as a solution for structural classification/clustering of XML documents.

Improving Collaborative Filtering Algorithms

2020 ◽  
pp. 638-657
Author(s):  
Firas Ben Kharrat ◽  
Aymen Elkhleifi ◽  
Rim Faiz
Keyword(s):  
Social Network ◽  
Web Sites ◽  
Opinion Mining ◽  
Semantic Analysis ◽  
Real Data ◽  
Data Sets ◽  
Suggested Approach ◽  
Opinion Score ◽  
Positive Results ◽  
Cold Start Problem

This paper puts forward a new recommendation algorithm based on semantic analysis as well as new measurements. Like Facebook, Social network is considered as one of the most well-prominent Web 2.0 applications and relevant services elaborating into functional ways for sharing opinions. Thereupon, social network web sites have since become valuable data sources for opinion mining. This paper proposes to introduce an external resource a sentiment from comments posted by users in order to anticipate recommendation and also to lessen the cold-start problem. The originality of the suggested approach means that posts are not merely characterized by an opinion score, but receive an opinion grade notion in the post instead. In general, the authors' approach has been implemented with Java and Lenskit framework. The study resulted in two real data sets, namely MovieLens and TripAdvisor, in which the authors have shown positive results. They compared their algorithm to SVD and Slope One algorithms. They have fulfilled an amelioration of 10% in precision and recall along with an improvement of 12% in RMSE and nDCG.

Sign in / Sign up

Export Citation Format

Share Document