mycourses .ntua.gr
Μοντέλα Οδηγούμενα από τα Δεδομένα σε Προβλήματα Μηχανικού

Γενικά στοιχεία

 

Περιγραφή

Δεδομένα από αισθητήρες (τύποι, χωρική και χρονική κάλυψη).
Κατηγοριοποίηση δεδομένων και ανάκτηση πληροφορίας-Δομές συσχέτισης.
Ανάλυση Fourier καi ανάλυση κύριων συνιστωσών (Principal Component analysis).
Ανάλυση δεδομένων από σταθερούς αισθητήρες. Ανάλυση δεδομένων από κινούμενους αισθητήρες.
Επεξεργασία δεδομένων από αισθητήρες κινητών τηλεφώνων (smartphone orientation, data cleaning, filtering, fusion, dimensionality reduction, feature engineering).
Πιθανοτικές μέθοδοι μηχανικής μάθησης τύπου Markov, Kriging, Polynomial Chaos etc.
Μοντελοποίηση με αναγωγή σε παραμετρικούς χώρους μειωμένης τάξεως (reduced order models). Εφαρμογές σε προβλήματα μηχανικού.
 
  • Teachinc hours 2
  •  Professors: Vissarion Papadopoulos, Eleni Vlahogianni, Michalis Fragiadakis
 
Ύλη

1

Smartphone sensing data

Technologies, data lifecycle, Preparing the data. Understanding the data. Modeling and evaluation. Actionable data driven models.

2

Data driven methods for smartphone sensing problems

Supervised learning, Unsupervised learning, Reinforcement learning. Theory-based Data Science Models.

3

Smartphone based analytics

Example I: Harsh event detection in R.

4

Understanding and modeling smartphone sensing data I

Example II: context aware detection problems in R.

5

Understanding and modeling smartphone sensing data II

Example III: User profiling in R.

6

From understanding to decision making

Example IV: Policy making with Reinforcement Learning, an application in R.

7

Simulation

Categories of simulation, Stochastic simulation, random numbers, Monte Carlo Simulation. Simple examples

8

Data extraction 

Data mining, autocorrelation, cross-correlation, power Spectrum. Discrete and continuous stochastic processes. Sampling and times series. White noise. Fourier transforms. Computational aspects of power spectrum. Analysis of large scale geophysical time series

9

Data driven models

Spectral representation and Karhunen–Loève series expansion methods. Simulation of stationary stochastic processes. Markov models

10

Data driven models

Reduced order models: Neural network, Support Vector Machines, Gaussian Processes (GPR, GPC).

11

Machine learning for structural performance assessment -1

Example 1: nonlinear response history analysis of a simple structural system

Example 2:  a rocking system

12

Machine learning for structural performance assessment - 2

Stochastic and Reliability analysis

Optimization under uncertainty.

13

Applications

Example: Reliability analysis of a simple structural system

Βιβλιογραφία

Bibliography

2. V. Papadopoulos and D. Giovanis, An Introduction to stochastic finite element methods, Springer (2017).

 

Bishop, C. M. (2006). Pattern recognition and machine learning. springer.

Chalmers, D. (2011). Sensing and systems in pervasive computing: Engineering context aware systems. Springer Science & Business Media.

Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep learning. MIT press.

Mέθοδοι αξιολόγησης

Performance evaluation

  • 50% Final exams
  • 50% Projects, homework and technical reports.
Διδάσκοντες

Professors:

Vissarion Papadopoulos

Eleni Vlahogianni

Michalis Fragiadakis

Βοηθοί

Dr. George Stavroylakis

Dr. Odysseas Kokkinos