Data Mining - Concepts and Techniques.

Data Mining Practical Machine Learning Tools and Techniques

As we are moving towards the Internet of Things (IoT), the number of sensors deployed around the world is growing at a rapid pace. Market research has shown a significant growth of sensor deployments over the past decade and has predicted a significant increment of the growth rate in the future. These sensors continuously generate enormous amounts of data. However, in order to add value to raw sensor data we need to understand it. Collection, modelling, reasoning, and distribution of context in relation to sensor data plays critical role in this challenge. Context-aware computing has proven to be successful in understanding sensor data. In this paper, we survey context awareness from an IoT perspective. We present the necessary background by introducing the IoT paradigm and context-aware fundamentals at the beginning. Then we provide an in-depth analysis of context life cycle. We evaluate a subset of projects (50) which represent the majority of research and commercial solutions proposed in the field of context-aware computing conducted over the last decade (2001-2011) based on our own taxonomy. Finally, based on our evaluation, we highlight the lessons to be learnt from the past and some possible directions for future research. The survey addresses a broad range of techniques, methods, models, functionalities, systems, applications, and middleware solutions related to context awareness and IoT. Our goal is not only to analyse, compare and consolidate past research work but also to appreciate their findings and discuss their applicability towards the IoT.

/pdf/context-aware-computing-for-the-internet-of-things-a-survey-3nibvvc921.pdf

Context Aware Computing for The Internet of Things: A Survey

Matrix Factorization Techniques for Recommender Systems

We quantitatively investigate how machine learning models leak information about the individual data records on which they were trained. We focus on the basic membership inference attack: given a data record and black-box access to a model, determine if the record was in the model's training dataset. To perform membership inference against a target model, we make adversarial use of machine learning and train our own inference model to recognize differences in the target model's predictions on the inputs that it trained on versus the inputs that it did not train on. We empirically evaluate our inference techniques on classification models trained by commercial "machine learning as a service" providers such as Google and Amazon. Using realistic datasets and classification tasks, including a hospital discharge dataset whose membership is sensitive from the privacy perspective, we show that these models can be vulnerable to membership inference attacks. We then investigate the factors that influence this leakage and evaluate mitigation strategies.

/pdf/membership-inference-attacks-against-machine-learning-models-2tzchejf91.pdf

Membership Inference Attacks Against Machine Learning Models

In this article, the authors present an approach for context-aware and QoS-enabled learning content provisioning, one of the essential elements in ubiquitous learning. The essence of the system is recommending the right content, in the right form, to the right learner, based on a wide range of user context information and QoS requirements. To facilitate knowledge interoperability and sharing, they modeled the learner context, content knowledge, and domain knowledge using ontologies. They first propose a knowledge-based semantic recommendation method to acquire the content the user really wants and needs to learn. Then, a fuzzy logic-based decision-making strategy and an adaptive QoS mapping mechanism determine the appropriate presentation according to user's QoS requirements and device/network capability.

/pdf/content-provisioning-for-ubiquitous-learning-2mh5dtn2kq.pdf

Content Provisioning for Ubiquitous Learning

Despite extensive research effort in contact-free sensing using RF signals in the last few years, there still exist significant barriers preventing their wide adoptions. One key issue is the inability to sense multiple targets due to the intrinsic nature of relying on reflection signals for sensing: the reflections from multiple targets get mixed at the receiver and it is extremely difficult to separate these signals to sense each individual. This problem becomes even more severe in long-range LoRa sensing because the sensing range is much larger compared to WiFi and acoustic based sensing. In this work, we address the challenging multi-target sensing issue, moving LoRa sensing one big step towards practical adoption. The key idea is to effectively utilize multiple antennas at the LoRa gateway to enable spatial beamforming to support multi-target sensing. While traditional beamforming methods adopted in WiFi and Radar systems rely on accurate channel information or transmitter-receiver synchronization, these requirements can not be satisfied in LoRa systems: the transmitter and receiver are not synchronized and no channel state information can be obtained from the cheap LoRa nodes. Another interesting observation is that while beamforming helps to increase signal strength, the phase/amplitude information which is critical for sensing can get corrupted during the beamforming process, eventually compromising the sensing capability. In this paper, we propose novel signal processing methods to address the issues above to enable long-range multi-target reparation sensing with LoRa. Extensive experiments show that our system can monitor the respiration rates of five human targets simultaneously at an average accuracy of 98.1%.

/pdf/unlocking-the-beamforming-potential-of-lora-for-long-range-sh2o21f8z8.pdf

Unlocking the Beamforming Potential of LoRa for Long-range Multi-target Respiration Sensing

With the developments in information and communications technology (ICT), people are involving in and connecting via various forms of communities in the cyber-physical space, such as online communities, opportunistic (offline) social networks, and location-based social networks. Different communities have distinct features and strengths. With humans playing the bridge role, these communities are implicitly interlinked. In contrast with the existing studies that mostly consider a single community, this article addresses the interaction among distinct communities. In particular, we present an emerging research area - cross-community sensing and mining (CSM), which aims to connect heterogeneous, cross-space communities by revealing the complex linkage and interplay among their properties and identifying human behavior patterns by analyzing the data sensed/collected from multi-community environments. The article describes and discusses the research background, characters, general framework, research challenges, as well as our practice of CSM.

Cross-community sensing and mining

Location prediction based on cellular network traces has recently spurred lots of interest. However, predicting one's location remains a very challenging task due to the randomness of the human mobility patterns. Our preliminary study included in this paper shows that there is a strong correlation and association among the certain group of users' locations. Through association pattern mining on Reality Mining dataset which involves 32,579 cell tower locations and 350,000 hours of continuous activity information, we observe the highly confident association rules exist among the locations of users, and then we further verify that the associations are indeed caused by the collective behaviors of the mobile phone users. Based on this finding we introduce the collective behavioral patterns (CBP), and then propose CBP-based predictor- a novel prediction schema that aims to forecasting one's locations in next 6 hours based on the locations of other users. Furthermore, we integrate the state-of-the-art i.e., Markov-based predictor with our CBP-based schema to build a hybrid predictor. We evaluate the CBP-based schema and compare the hybrid predictor with the Markov-based predictor through intensive experiments. Experimental results show that CBP-based predictor achieves good precision and the hybrid predictor produces higher prediction accuracy than the state-of-the-art scheme at cell tower level in the forthcoming one to six hours. Finally it is verified that collective behavioral patterns can be used to predict user locations as well as to improve the performance of existing predictors.

Predicting Mobile Phone User Locations by Exploiting Collective Behavioral Patterns

Providing accurate information about human's state, activity is one of the most important elements in Ubiquitous Computing. Various applications can be enabled if one's state, activity can be recognized. Due to the low deployment cost, non-intrusive sensing nature, Wi-Fi based activity recognition has become a promising, emerging research area. In this paper, we survey the state-of-the-art of the area from four aspects ranging from historical overview, theories, models, key techniques to applications. In addition to the summary about the principles, achievements of existing work, we also highlight some open issues, research directions in this emerging area.

Daqing Zhang

Papers

Content Provisioning for Ubiquitous Learning

Unlocking the Beamforming Potential of LoRa for Long-range Multi-target Respiration Sensing

Cross-community sensing and mining

Predicting Mobile Phone User Locations by Exploiting Collective Behavioral Patterns

A Survey on Wi-Fi Based Contactless Activity Recognition