Présentation EPFL-Public | Ecole Polytechnique Fédérale de Lausanne EPFL
Présentation EPFL-Public | LSIR Projects Goal: Develop a participatory platform towards the conceptualization of scientific knowledge and its large scale analysis. Achievements: Developed a web-based system already used by many scientists for semantically importing, storing and searching scientific data. Obtained a crowdsourced ontology with over 15K concepts for physics and life science Motivation: The volume of research information outgrows our individual capacity to process it Publications and knowledge are scattered across different sources Scientific achievements rely on multiple disciplines ScienceWise – JRA1 A web-based scientific platform to boost research
Présentation EPFL-Public | LSIR Projects CloudSpaces –JRA5 Open Service Platform for the Next Generation of Personal Clouds Easily usable privacy enhancing tools for end-users. Also allow users to transparently access data stored at multiple cloud services Allow small datacenters to share their resources to meet user demand For privacy, we have proposed a novel framework that relies not just on content but also the context or semantics of data sharing We base our framework on psychologically grounded models in order to determine privacy preferences We use Crowdsourcing for determining sensitivity of sharing operations Our software provides policy recommendations to users in order to share their files safely on any cloud platform
Présentation EPFL-Public | LSIR Projects Motivation: Internet has become the primary source of information Misleading, questionable and factually incorrect Web pages Serious consequences for decision-making in critical domains Achievements: Novel Web credibility evaluation mechanisms by leveraging crowdsourcing, big data analytics, machine learning, etc. Game theoretic models for robust credibility evaluation. A unique large-scale credibility corpus Web credibility evaluation prototype system ( Use cases: Credible Web search Online medical information credibility evaluation Credibility control in an open shared knowledge base ( Goal: Creation of new mechanisms for supporting users in the evaluation of Web content credibility. Machine learning based credibility assessment Snapshot of Reconcile credibility evaluation system prototype Reconcile – JRA5 Robust Online Credibility Evaluation of Web Content
Présentation EPFL-Public | LSIR Projects Providing incentives to consumers to alter their behaviour and ensure energy balance Observing energy consumption data to infer patterns and measure the influence of DR on energy consumption Approach: Energy consumption disaggregation, prediction and segmentation using statistical machine learning methods Game-theoretic, economic and behavioural models of energy consumers Field trials with residential and commercial consumers Results: Design guidelines for effective incentives to change energy consumption patterns Tools and algorithms for energy data analytics Wattalyst –JRA8 Modelling and Analysing Demand Response (DR) Systems