Search

«Νέες τεχνολογίες και ρομποτική στην Έρευνα και Διάσωση» #SearchAndRescue #ICT #Robotics #AI #lecture Περιεχόμενα: Επιχειρήσεις Έρευνας και Διάσωσης σε αστικό περιβάλλον / προβλήματα και τεχνολογικά «κενά» / Άξονας 1: τεχνολογίες για την ασφάλεια του διασώστη / Άξονας 2: τεχνολογίες εντοπισμού θυμάτων / Άξονας 3: τεχνολογίες τηλεπικοινωνιών και συντονισμού / «έξυπνες» ρομποτικές τεχνολογίες για έρευνα σε ερείπια / Μελέτη περιπτώσεων από ερευνητικά προγράμματα EU Horizon 2020. Επίπεδο γνώσεων (από): Οποιοδήποτε Προαπαιτούμενα: Κανένα &nbsp

«Εφαρμογές της Τεχνητής Νοημοσύνης στον πραγματικό κόσμο» Περιεχόμενα: Κατηγοριοποίηση προβλημάτων στον πραγματικό κόσμο / συνήθεις τρόποι επίλυσης / «δύσκολα» προβλήματα που απαιτούν προσεγγιστικές λύσεις / ο ρόλος της Μηχανικής Μάθησης και της Τεχνητής Νοημοσύνης στην επίλυσή τους / μελέτη περίπτωσης: Αυτόματη ιατρική διάγνωση στη μαστογραφία. Επίπεδο γνώσεων (από): Μαθητές Λυκείου και φοιτητές Προαπαιτούμενα: Βασικές γνώσεις Μαθηματικών και Στατιστική&sigmaf

«Εισαγωγή στη Μηχανική Μάθηση και στην Αναλυτική Δεδομένων» Περιεχόμενα: Βασικές έννοιες Τεχνητής Νοημοσύνης, Μηχανικής Μάθησης και Αναλυτικής Δεδομένων / συλλογή, προεπεξεργασία και ανακατασκευή δεδομένων / σχεδίαση ροής (pipeline) / επιλογή χαρακτηριστικών (features) / εκπαίδευση μοντέλων με επίβλεψη (classification, regression) και χωρίς επίβλεψη (clustering, associations) / απεικόνιση δεδομένων (visual analytics). Επίπεδο γνώσεων (από): Μαθητές Γυμνασίου – Λυκείου Προαπαιτούμενα: Στοιχειώδεις γνώσεις Μαθηματικώ&nu

Water resource management is one the most urgent aspects of environmental protection and sustainability policies world-wide. Accurate, real-time remote sensing of the status of underground reservoirs is required for proper regional planning, prevention of droughts, optimized farming etc. ‘Water Underground’ is a low-cost solution, based on a combination of Internet of Things (IoT) local sensing, Edge computing, Cloud storage, web services and Machine Learning (ML) and predictive analytics, continuously monitoring the level of underground water and its quality. Specifically, water level is monitored via an IoT apparatus providing the Static (SWL) and Pumping Water Level (PWL). Moreover, the quality of water is tracked via measuring the Total Dissolved Solids (TDS), Oxidation-Reduction Potential (ORP), temperature, pH, electrical-conductivity, etc. Local processing in the IoT device includes measurements’ transformations and robust adaptive control for the device’s actuators. The reservoir dynamics is tracked and modeled using Cloud-based predictive analytics. The corresponding Cloud services include long- and short-term detection of periodic trends, Drawdown (DD) patterns, prediction of SWL, predictive maintenance via PWL tracking, etc. The overall solution has received international recognition in IBM Challenge 2020 as top-7 finalist for Europe. The platform is currently under prototype deployment in several sites in the Attica region of Greece.

Flood risk prediction requires consistent and accurate sensor measurements, usually provided from traditional in-situ environmental monitoring systems. Crowd-sourced data can complement these official data sources, allowing authorities to improve and fill gaps in the hazard assessment process. However, collecting this information from volunteers, with no technical knowledge and while using low-cost equipment such their smartphones and tablets, raises the question of quality and consistency. To alleviate this barrier two tools were developed in the context of H2020 Scent project. The Water Level Measurement Tool uses image recognition techniques to extract the water level from images containing a measuring tape. The Water Velocity Calculation Tool uses video processing algorithms to extract the water surface velocity from a video containing a pre-defined floating object moving on the surface of a water body. Each extracted measurement is accompanied by a degree of trust. The tools have been designed so that a high degree of trust can be achieved from images and videos taken from regular smartphones. The crowdsourced river measurements are used to develop improved flood models with a dramatically reduced cost as both the measuring tapes and the floating object are low-cost and re-usable while effectively covering large areas of interest.

This is a short review and commentary on a paper that was recently published, regarding the effects of opening the borders during the summer of 2020 in Greece and how it allegedly did not result in significant SARS-CoV-2 imported infections from travellers. In this commentary, this thesis is proven unsupported by sufficient evidence, as the study presents several deficiencies and biases regarding the statistical context and the proof of the core hypothesis, as described herein., This report is not supported by nor associated with any research project or funding grant.

This is a short review and commentary on a paper that was recently published, regarding the employment of an Artificial Intelligence system for targeted testing of travellers for SARS-CoV-2 at the border checks during the summer of 2020 in Greece. The core thesis of the paper is that the system offered significant accuracy and risk mitigation mechanisms for this task. However, the methodology and the supportive material presented therein shows several deficiencies and biases that do not properly justify such claim. The main problems with the paper are presented here, with short discussion about how they should be addressed by the authors.

Since the beginning of 2020, COVID-19 is the most urgent and challenging task for the international scientific community, in order to identify its behaviour, track its progress and plan effective mitigation policies. In this study, Greece is the main focus for assessing the national outbreak and estimating the general trends and outlook of it. Multiple data analytics procedures, spectral decomposition and curve-fitting formulations are developed based on the data available at hand. Standard SEIQRDP epidemic modelling is applied for Greece and for the general region around it, providing hints for the outbreak progression in the mid- and long-term, for various infections under-reporting rates. The overall short-term outlook for Greece seems to be towards positive, with a downward trend in infections rate daily increase (i.e., now beyond the exponential growth rate), a possible peak within a few days beyond April 14th, as well as the high availability level of ICU w.r.t. expected demand at peak. On the negative side, the fade-out period seems to be in the order of several months, with high probability of recurrent surges of the outbreak. The mitigation policies for the `next day' should be focused on close tracking of the epidemic via large-scale tests, strict border checking in international travelling and an adaptive plan for selective activation of mitigation measures when deemed necessary. Keywords: COVID-19, SARS-CoV-2, data analytics, SEIR, Greece, Additional data sources and LSE solver details over the previous version 1.2.00 / 17-Apr-2020.

Octave, R and Python identical codes are tested in terms of in terms of end-user execution speed, using a very low-end "embedded" hardware system and a standard office workstation. The codes include algorithmic primitives common in Data Analytics and Machine Learning, i.e., matrix manipulation (inversion, product), linear Algebra, linear regression, Singular Value Decomposition (SVD), fast Fourier transformation (FFT) and a baseline Bubblesort implementation for testing flow control structures. Description In Data Analytics and Machine Learning, code prototyping is an integral part of the Research & Development (R&D) process, especially in data exploration and algorithm design. The programming tools and platforms used for these tasks are selected for rich API/library base, high-level expression syntax, very compact code, interactive on-the-fly code input, abstract data management and best-possible execution speed. Thus, traditional programming languages are usually inappropriate for such heavily iterative and exploratory coding evolutions. Today, by far the three most popular and appropriate choices are Octave, R and Python. In this work, these three programming environments are assessed in terms of end-user execution speed. More specifically, some common algorithmic primitives are implemented and tested in each language separately, including matrix manipulation (inversion, product), linear Algebra, linear regression, Singular Value Decomposition (SVD), as well as fast Fourier transformation (FFT) as a standard procedure in a signal processing pipeline. Additionally, a baseline implementation of the Bubblesort algorithm is employed for testing the efficiency of flow control structures and execution performance in code branching. The results present the performance of the three identical source codes in terms of end-user execution speed (elapsed time) in three different hardware platforms, namely: (1) simulating very low-end processing and resources machine similar to embedded systems (Linux, 2GB RAM, N20 Atom single-core CPU), (2) a standard/enhanced office workstation (Win10, 16GB RAM, dual-core i7 CPU) and (3) a high-end workstation or small office server (Win10, 32GB RAM, quad-core i7 CPU)., Conference: FossComm 2018 @ 13-14 October, Heraklion, Greece.

Aircraft trajectory prediction (TP) is a challenging and inherently data-driven time-series modeling problem. Adding annotation parameters further increases the complexity of the search space, especially when ‘blind’ optimization algorithms are employed. In this paper, flight plans, localized weather and aircraft properties are introduced as trajectory annotations (or semantics), which enable modeling in a space higher than the typical 4-D spatio-temporal domain. A two-phase hybrid approach is employed for the core TP task: (a) clustering using properly designed semantic-aware similarity functions as distance metrics; and (b) a hidden Markov model (HMM) for each cluster, using non-uniform graph-based spatial grid and exploiting flight plans as constraints for a parametric probabilistic model for the emissions. The proposed method is applied in real radar tracks and weather data for a one-month dataset of flights in Spanish airspace. Using parametric Gaussians as the base for the emissions model and confidence interval estimations for the associated errors, the proposed method exhibits exceptionally low HMM complexity and per-waypoint prediction accuracy of a few hundred meters compared with submitted flight plans., Intermediate technical report for work-in-progress (Oct/2017).

In this study, the wind data series from five locations in Aegean Sea islands, the most active `hotspots' in terms of refugee influx during the Oct/2015 - Jan/2016 period, are investigated. The analysis of the three-per-site data series includes standard statistical analysis and parametric distributions, auto-correlation analysis, cross-correlation analysis between the sites, as well as various ARMA models for estimating the feasibility and accuracy of such spatio-temporal linear regressors for predictive analytics. Strong correlations are detected across specific sites and appropriately trained ARMA(7,5) models achieve 1-day look-ahead error (RMSE) of less than 1.9 km/h on average wind speed. The results show that such data-driven statistical approaches are extremely useful in identifying unexpected and sometimes counter-intuitive associations between the available spatial data nodes, which is very important when designing corresponding models for short-term forecasting of sea condition, especially average wave height and direction, which is in fact what defines the associated weather risk of crossing these passages in refugee influx patterns., Ref.No: HG/DA.0725.01v1 – Licensed under Creative Commons (BY-NC-SA) 4.0 - Copyright (c) 2017 Harris Georgiou.

Functional Magnetic Resonance Imaging (fMRI) is a powerful non-invasive tool for localizing and analyzing brain activity. This study focuses on one very important aspect of the functional properties of human brain, specifically the estimation of the level of parallelism when performing complex cognitive tasks. Using fMRI as the main modality, the human brain activity is investigated through a purely data-driven signal processing and dimensionality analysis approach. Specifically, the fMRI signal is treated as a multi-dimensional data space and its intrinsic `complexity' is studied via sparsity-promoting matrix factorization in the sense of blind-source separation (BSS). One simulated and two real fMRI datasets are used in combination with Independent Component Analysis (ICA) for estimating the intrinsic (true) dimensionality via detection of statistically independent concurrent signal sources. This analysis provides reliable data-driven experimental evidence on the number of independent active brain processes that run concurrently when visual or visuo-motor tasks are performed. The results prove that, although this number cannot be defined as a hard threshold but rather as a continuous range, however when a specific activation level is defined, an estimated number of concurrent processes or the loose equivalent of `brain cores' can be detected in human brain activity., Preprint version, final appears in: Panhellenic Conference on Informatics (PCI2017), September 28-30 2017, Larisa, Greece (ACM in ICPS Series).

During the last two years, hundreds of thousands of refugees have traveled across the Mediterranean Sea, in perilous conditions and unsafe boats, resulting in thousands of dead and missing, despite the best rescue efforts from both sides. Data analysis and statistics may have an answer for the challenging task of predicting the flows of incoming boats and assisting in the prompt allocation of Search & Rescue (SAR) resources at the frontlines.

In this paper, the optimality of ternary arithmetic is investigated under strict mathematical formulation. The arithmetic systems are presented in generic form, as the means to encode numeric values, and the choice of radix is asserted as the main parameter to assess the efficiency of the representation, in terms of information compactness and estimated implementation cost in hardware. Using proper formulations for the optimization task, the universal constant e (base of natural logarithms) is proven as the most efficient radix and ternary is asserted as the closest integer choice.

Advanced sensing technologies, combined with wireless sensor networks, have already demonstrated their value in monitoring urban water systems, where management is rather centralized within water utility organizations. Environmental water resources, characterized by more diverse stakeholders and overlapping management responsibilities of different agencies, present more challenging contexts for implementing novel sensing technologies. Crowdsourcing by citizens has been proposed as an alternative approach for adaptive data collection that can augment the amount of data collected, as well as bring together the diverse stakeholders and citizens in more participatory water resources management processes. This article first introduces the challenges of designing citizens' campaigns for collecting data on environmental waters. A set of developed mobile phone and web applications is then introduced, integrated within a specific platform, as it was used in the execution of citizens' campaigns for data needed in flood analysis and management. Experiences and lessons learned are presented from the field execution of citizens' campaigns in two pilot areas located in Europe - the Danube Delta in Romania, and the Kifissos catchment in Greece. Two of the campaigns are on river data collection - water levels and water velocities, and two on collecting land use/land cover data. Surveys carried out with campaign participants indicate their appreciation of the initiative, but challenges remain regarding user-friendliness of the applications. Logistic issues such as timing, duration, and pathways for data collection impacted the motivation of participants. Overall, a unique and large dataset was obtained in terms of quantitative water measurements, despite data losses due to low raw data quality. Further work lies in evaluating the usability of this dataset for local authorities.

Innovative technologies can enhance operational capabilities of First Responders (FRs) during Search & Rescue (SAR) operations, while at the same time increasing safety levels. The INGENIOUS project1 (EU Horizon 2020) aims at developing, integrating, testing and validating a next generation SAR toolkit for collaborative response, which ensures high level of protection and augmented operational capacity in disaster situations. In this paper, a subset of components of the toolkit mostly focused on increasing situational awareness and safety are described: the Fusion Engine (FE), which receives data from multiple sources, stores and analyzes them for integration purposes regarding situational awareness and sends them to the Common Operational Picture (COP) as well as the mobile FR terminals; the Smart Boots (SB), which collect data of individual FRs and provide information regarding their health status and alerts; the Modular Airborne Camera System (MACS); the Multi‐purpose Autonomous eXploring (MAX) drone for indoor/outdoor mapping and assessment of unknown environments; and the Micro INdoor drones (MINs) used for FRs indoor localization. The functionalities of these components, as well as the first prototypes developed and currently under lab and field test, are presented in this paper.

The tremendous growth of positioning technologies and GPS enabled devices has produced huge volumes of tracking data during the recent years. This source of information constitutes a rich input for data analytics processes, either offline (e.g. cluster analysis, hot motion discovery) or online (e.g. short-term forecasting of forthcoming positions). This paper focuses on predictive analytics for moving objects (could be pedestrians, cars, vessels, planes, animals, etc.) and surveys the state-of-the-art in the context of future location and trajectory prediction. We provide an extensive review of over 50 works, also proposing a novel taxonomy of predictive algorithms over moving objects. We also list the properties of several real datasets used in the past for validation purposes of those works and, motivated by this, we discuss challenges that arise in the transition from conventional to Big Data applications.