You're about to create your best presentation ever

Phd Defense Presentation Template

Create your presentation by reusing one of our great community templates.

PhD defense

Transcript: Partial vs. impartial Trust Personality Competence A need for the question: who performs the practice we want to change? Gender focus in practical initiatives and policy programmes 4 social dimensions When you have received the text messages to change the timing of your electricity consumption, how often have you decided to do the following in order to change the timing of your electricity consumption? Changed the timing of washing (1)  (2)  (3)  (4)  (5)  (6)  Changed the timing of dishwashing (1)  (2)  (3)  (4)  (5)  (6)  Changed the timing of tumble drying (1)  (2)  (3)  (4)  (5)  (6)  Changed the timing of charging of phone, etc (1)  (2)  (3)  (4)  (5)  (6)  "When we bought the house, it had a thatched roof. It was old and needed to be changed. We could easily have chosen another type of roof. It would have been cheaper and cost less in insurance. But, no, it had to be a thatched roof. We wanted to be faithful to the original style of the house, which is why we also kept the old plank flooring". never Background INCAP Social and gendered engagements significantly impact decisions on energy renovations and practices of flexible electricity use. Aim: to investigate the importance of gendered practices for how households react to flexible electricity use. West and Zimmerman This is your area, John Motivation of the private households Focus on collective action Reaching the environmental sceptics More power and security Relying on a few passionate participants Little experience in the area Lise Tjørring, Industrial PhD defense, November 16th 2017 Partners The positioning in the life cycle influenced people's willingness and capacity to energy renovate Targeting women or rethinking the way we develop solutions? Loosening the boundaries of fixed categories? Financial incentives vs. a focus on (gendered) practices Theoretical attention Aim: to investigate the significance of gender in private households for decisions on energy renovations. Those green tiles in the bathroom, they have to go Foucault From homogenous households to engagements between household members, home and energy advisor Conducted by co-author: • Participant observation in a variety of meetings, an energy fair and conferences • Literature study • Semi-structured interviews with project managers, citizens and builders The process of analysis I'm in the bathroom for 5 min Research focus Paper 1 addresses the social engagement between the energy advisor and the family and between the home and the human actors in the context of energy renovation. Paper 2 addresses the gendered engagement between the energy advisor and the family in the context of energy renovation. Paper 3 addresses the gendered engagement between the energy advisor and the family in the context of flexible electricity use. Paper 4 addresses the social engagement between the home and the human actors in the context of energy renovation. Austin A gap between what the energy advisors think people want and what people really want The importance of empathy Homes instead of houses Influencing vs. being influenced Data from interviews Paper 2: Conclusion How important will it be for you to move the timing of your electricity consumption when you, during the test period, receive text messages telling you the best time of the day to use electricity? (1)  Very important (2)  Important (3)  Slightly important (4)  Less important (5)  Not important (6)  Don’t know Bias Data collection As perceived by the households Aim: An investigation of the possibilities for flexible electricity consumption in private households. Partners: University of Copenhagen, the energy company SE, the Technical University of Denmark (DTU), Stanford University and the private company Develco. One year field experiment with 71 participants. Main findings Althusser We both retire this summer, then we'll get the time and money to renovate......... Merleau-Ponty Paper 1: Energy renovation models in private households in Denmark Research objective "I often answer a question twice. First in a technical manner, and then I talk about the soft values..." Phenomenological anthropology Practical engagement in the world The dwelling perspective I don't want people to drive through the village and see my ugly solar panels The role of the energy adviser Financing aspects Project evaluation methods The value of community vs. individual approaches Motivation of the private households Questionnaire 1 Questionnaire results Money Comfort Task Man Woman Both Laundry 10% 70% 20% H.M. 100% Cooking 10% 50% 40% Cleaning 10% 90% Dishes 20% 20% 60% not very often The typical energy consultation: - the energy advisor primarily talks to the man - the man takes on an active role, the woman a passive role - Taking for granted that the man knows about maintenance Men and women have different priorities for energy renovation based on their practices in the house The connection between gender and doings Practice as a

PhD defense

Transcript: 1- Introduction 2- Theories and method 3- Chaterization and application 4-Conclusion 5- The connection Drug Delivery: -What ? -Why ? * The release ability of these two materials have been compared in term of using small molecules drug. However, silica nanotubes proved to be more suitable with large molecules drug. * This nanostructured tubes have been used with anti-cancer Doxorubicin and optimized the geometry for minimum energy using Hyperchem. Resultes were comparable with previously reported data. * Steric effects plays an important role in the loading efficency. Si-Ti-Sv * The efficiency of silica-titania sieves as bioactive materials as well as their potential to be a suitable nano-carrier have been examined by using a new anti-bacterial agent, Izohidrafural * Izohidrafural, the new antibiotic agent exhibited the highest antimicrobial efficiency, superior to the common drug Nitrofurantoin against the majority of E. coli strains, with average MIC of (4.68 µg/mL), followed by silica-titania sieves loaded with Izohidrafural. * The non-loaded silica-titanium sieves exhibited the highest antimicrobial activity against the gram-positive cocci. Nitrofurantoin Preparation technique Sol-Gel Sol-Gel Adjustments: Synergistic effect The synergistic effect might occur due to the fact that the Titanium dioxide, in form of anatase, in Si-Ti-Sv, has photocatalytic properties and thus the ability of creating reactive oxygen species (ROS)... Reported mechanism.G. Timmins et. al. (2006) The biological activity of isoniazid states (in Izo.) The presence of an oxidizing enzyme, KatG, the molecule is transformed into radical species BY TiO2 such as acyl, acylperoxo and pyridyl radicals... We may presume that TiO2 plays a role in activating Izohidrafural producing thus a synergistic effect... http://www.sigmaaldrich.com/technical-documents/articles/material-matters/mesoporous-materials.html Nanostructured silica MCM-41 The chemical structure for this drug is analogue to some extent with Nitrofurantoin, Urinary Tract Infections On the other hand the ORIENTATION of the calculated dipole moment, YELLOW ARROW favors the penetration of the molecule in polar cavities as it is for MCM-41. = slightly Higher Loading http://www.chromacademy.com/lms/sco5/10-Selecting-Reversed-Phase-Columns.html fChannel=6&fCourse=64&fSco=410&fPath=sco5/10-Selecting-Reversed-Phase-Columns.html Nitrofurazone Quantum Mechanics Izohidrafural Ibuprofen: Anti-inflammatory drug BASE MATERIAL -Adding Surfactant This method will be applied with other drugs For Detecting EE % Other Characterizations: *Three types of silica nanostructured materials have been prepared as drug carriers *The preparation technique follows the principals of sol-gel method, where the prepared silica materials were MCM-41, silica nanotubes and silica-titania sieves. *The obtained nanomaterials were characterized according to different considerations *The characterization methods were: SEM, TEM, EDAX, XRD, BET and FT-IR. *Silica-titania six different samples, according to different preparation medium and calcination conditions. *Silica nanotubes and MCM-41 were compared as a carrier materials for Ibuprofen with different encapsulation efficiencies and release profiles . The reason behind that proved to be related to the pore size of those two materials. General Atomic and Molecular Electronic Structure System (GAMESS) FTIR REFERENCES [1] L. P. Singh, S. K. Agarwal, S. K. Bhattacharyya, U. Sharma, S. Ahalawat, Nanomater. nanotechnol. 1 (2011) 44-51. [2] R. Stan, C. Ott, S. Rosca, A. Badanoiu, S.Stoleriu, G. Voicu, U.P.B. Sci. Bull., Series B 3 (2008) 1454-2331 [3] F. Kamar, A. Nechifor, M. Ridha, M. B. M. Al Tameemi, G. Nechifor, REV. CHIM. 66 (2015) 921-925. [4] Fruijtier-Polloth C., Toxicology 294 (2012) 61– 79. [5] D. Dykxhoorn, J. Lieberman, Cell 2 (2006) 231-5. [6] D. Dykxhoorn, D. Chowdhury, J. Lieberman, Dordrecht: Springer (2008) 299-329. [7] M. B. M. Al tameemi, Master thesis, EMU, university publications, Doi: http://hdl.handle.net/11129/1517 [8] L. Yan, X.Chen, Nanocrystalline Materials (2013) 221-268. [9] S. Ranghar, P. Sirohi, P. Verma, V. Agarwal, Braz. Arch. Biol. Technol 57 (2014) 209-222. [10] C. S. S. Kumar, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim (2006). ISBN: 3-527-31382-6 V [11] C. S. S. Kumar, WILEY-VCH Verlag GmbH & Co. KGa (2007). [12] I. Rahman, V. Padavettan, Journal of Nanomaterials (2012) 15. [13] W. Stober, A. Fink, J. Bohn, J. Coll. Int. Sci. 26 (1968) 62. [14] G. Bogush, C. Zukoski, J. Non‐Cryst. Sol. 104 (1988) 95. [15] J. Shi, L. Guo, F. Cui, X. Cui, Microporous and Mesoporous Materials 117 (2009) 609. [16] I. Rahman, P. Vejayakumaran, C. Chee, Ceramics International 32 (2006) 691. [17] Z. Wang, J. Zhao, X. Ding, K. Yu, Materials letters 59 (2005) 4013. [18] K. Ikari, K. Suzuki, H. Imai, Langmuir 22 (2006) 802. [19] Y. Han, J. Ying, Angewandte Chemie International Edition 44 (2004) 288–292. [20] T. Tan, S. Liu, Y. Zhang, M.-Y. Han, S. Selvan,

PhD defense

Transcript: Outlook bead radius viscosity excellent agreement of theory and experiment no additional terms needed symmetry breaking necessary for net pumping of fluid rows, arrays of cilia: larger pumping effects, nearly homogeneous flow above the ciliated surface external magnetic field t = 160 s opening angle, tilt angle, offsets, adding the microparticles, initial distribution vacuum permeability ability of support, adhere the cells, promotion of cell growth a variety of possible scaffold shapes, possibility of enlargement during experiment low Reynolds number hydrodynamics - magneto-optical tweezers - self-assembled chains composed of superparamagnetic beads - self-assembled colloidal scaffolds wave optics description Navier-Stokes equation point dipole wavelength Gašper Kokot adviser: dr. Andrej Vilfan SRA adviser: dr. Mojca Vilfan Hydrodynamic coupling marked difference between longitudinal and transverse coupling coefficients for rods (not 2 as in bulk for a sphere) - wall effects distance dependence of the coupling coefficient of the rods was found to be different for longitudinal and transverse direction - models have to take it into account brightfield optical tweezers pulling fluorescence magnetic susceptibility biomimetics superparamagnetic beads simulation: mobility matrix with the Rotne-Prager approximation incompressibility Measuring forces with magneto-optical tweezers in biological and biomimetic systems magnetic field calibration volume Self-assembled artificial cilia Aim artificial system: simplified, more control over parameters physical model with external drive Introduction 1/d dependance for both longitudinal (purple) and transverse (blue), ratio stays 2 cell growth proof of principle density magnetic filed, no colloids swimming, pumping, feeding, cleaning, body symmetry determination ... magneto-optical tweezers detailed study of flow around a cilium direct illumination photolithography ray optics description Colloidal scaffolds for cell growth clean glass external force density t = 0 s typical velocity bead separation (scaffold bearing force measurements) t = 20 s efficient biological systems typical length evolution pressure artificial cell support: self-assembled structures artificial cilia: microfluidic pumps collective effects: coordinated movement rapid magnetic field oscillation 3 velocity Stokes equation hydrodynamic coupling coefficient measurements development of a simple method of artificial cilia carpets fabrication studying other beating patterns hydrodynamic coupling under combined influence of several cilia heterostructured particles - a variety of shapes and colloids colloidal scaffolds - coatings, interaction potentials, shapes, sizes slow oscillation of magnetic field resistive force theory, Blake tensor expansion to 1/r terms, temporal and spatial average: light sensitive polymer cross-linking after exposure to UV light hydrodynamic coupling measurement: driven and passive cilium t = 120 h Experimental methods cilia: arrays, complicated beating pattern fluorescent tracer particles trajectory analysis flow maps pumping velocity calculated colloids, no magnetic field Actuated flow

PhD Defense

Transcript: Reaction Convection Diffusion Thank you for your attention Pressure in the stented arterial wall Matching conditions on the interface Biodegradable Polymer (PLA) Fully embedded stent in the arterial wall Stiffness of the arterial wall Affinity of the arterial wall Future works Evolution of the drug mass in the arterial wall for different values of κ_r for long time (Maxwell-Wiechert model) Center for Mathematics, University of Coimbra (CMUC) FCT-Grant SFRH/BD/51167/2010 Acknowledgments Velocity and pressure in the arterial wall 3-parameter solid model Tiny expandable metallic mesh tube Non-Fickian Models for Biodegradable Drug Eluting Stents Convection Weak formulation Numerical study Modeling Where Velocity in the stented arterial wall Main assumptions PDE solver freeFEM++ 3688 elements (1968 vertices) for the arterial wall and 100 elements (83 vertices) for each stent Implicit-Explicit (IMEX) backward formula in the uniform time grid Piecewise linear finite element space for concentration and pressure Velocity and pressure in the stent A DES with topcoat Reaction Conclusions and future works Numerical Experiments Coupling the model with the blood flow equations Gradual stent embedding using moving boundary problems Fully bioabsorbable stent (ABSORB, Abbott Vascular) DES without topcoat Analytical study Mathematical modeling Qualitative behavior of the total mass of the system Motivation Maxwell-Wiechert viscoelastic model Conclusions 2D coupled model to simulate drug release from a polymeric stent to the arterial wall. Biodegradable coating and viscoelasticity and affinity of the arterial wall Three particular aspects of clinical importance are included: A: The influence of the stiffness of the arterial wall on the drug accumulation in the arterial wall, B: The effect of permeability of the stent coating on the drug release from the stent, C: The effect of affinity of the arterial wall on the dug release, Atherosclerosis Stability Drug Eluting Stent Drug release from polymeric stent into the arterial wall Publications where Viscoelastic effect Ferreira, J.A., Naghipoor, J. and de Oliveira, P. (2015), Numerical and analytical study of a coupled cardiovascular drug delivery model, Journal of Computational and Applied Mathematics, 275, 433-446 Ferreira, J.A., Naghipoor, J. and de Oliveira, P. (2013), Numerical simulation of a coupled cardiovascular drug delivery model, Proceeding of 13th International Conference on Computational and Mathematical Methods in Science and Engineering, , CMMSE2013 (II), I. P. Hamilton and J. Vigo-Aguiar (editors), 642-653, Almeria, Spain Ferreira, J.A., Naghipoor, J. and de Oliveira, P. (2014), The effect of reversible binding sites on the drug release from drug eluting stent, Proceeding of 14th International Conference on Computational and Mathematical Methods in Science and Engineering, CMMSE2014(II), I. P. Hamilton and J. Vigo-Aguiar (editors), 519-530, Cadiz, Spain Ferreira, J.A., Naghipoor, J. and de Oliveira, P., A Coupled Non-Fickian Model of a Cardiovascular Drug Delivery System, CMUC Preprint 14-13, Submitted Evolution of the drug mass in the arterial wall for different values of κ_r for long time (Fung’s model) T. Khamdaenga, J. Luo , J. Vappou, P. Terdtoon and E.E. Konofagou, Arterial stiffness identification of the human carotid artery using the stress-strain relationship in vivo, Ultrasonic 52 (3) (2012) 402-411. Arterial stiffness has been shown to be an excellent indicator of cardiovascular morbidity and mortality in a large percentage of the patients with atherosclerosis Chemical reactions in the stent 1) 1 2) 1 2 Chemical reaction in the vessel wall Reversible binding in the vessel wall Reactions in the coated stent Reactions in the arterial wall + initial and boundary conditions Initial mass the mass of plasma that enters in the system until time t the mass of hydrolyzed oligomers until time t the mass of the components that are on the boundary until time t Distribution of drug in the arterial wall in the model without binding, 30 days Distribution of drug in the arterial wall in the model with binding, 30 days

PhD Defense

Transcript: Metrewave HDR Snapshot Spectroscopic imaging studies of weak coronal events Atul Mohan Supervisor: Divya Oberoi Collaborators: Surajit Mondal, Patrick McCauley, Alpha Mastrano Science drive & Methodology Introduction Solar corona is a tenuous layer of plasma at ~ 2 MK , Bright in EUV, X-ray m-wavelengths (~ 100 MHz). Plasma instabilities coherent emission General Properties: 211 Fe XIV ~ 2 MK Introduction Weak Energy release events: Teeming with weak energy release events. Dominate flare energy histograms (Crosby et al., 1993; Mercier & Trottet 1997; Aschwanden 2011) 37 yrs of X-ray flare obs. ~ 90% microflares ~ 1 event / hr (Aschwanden & Freeland 2012) ~ 56 events/min !! Suresh,A et al., 2017 Scientific relevance of weak events A major source of local coronal heating Weaker cousins of strong Earth impacting events (Moore et al.,2010, Hong et al., 2011, 2013) Don't alter physical fields allows to better constrain the mechanism. Large no. ensemble of possible scenarios of "heating & particle acceleration" But!! Not well understood & most studies are HE based Challenges to study HE thermal emission: Weak signatures. Low energy ones unreported in event lists Coherent emission at m-bands easy to detect. Multiple weak bursts occur simultaneously. Need DR ~ 1000 to track continuously Rapid spectro-temporal variation ( < s, ~ 10 kHz) Need high DR spectroscopic snapshot imaging + HE imaging + magnetogram based study. 0.5, 1, 3, 5, 10, 30, 60, 90 % of peak Good news - HDR Snapshot spectroscopic imaging is here !!! MWA - 128 tiles in ~ 3 km diameter, 8128 baselines Dense instantaneous uv sampling MWA is best suited. ( 0.5 s & 40 kHz resolution ; 80 - 300 MHz band) Mondal et al, 2019 New Challenges: Data Volume. Need for automated imaging & calibration pipelines. Tools & Techniques Big data problem Image Calibration 4D DataCubes & SPREDS Part I: Tools & Techniques Big data Introduction MWA data output rate: 1 TB/hr In Pawsey Supercomputing Centre, Australia: 3000 hrs of data Download speed from repo: ~ 1 GB/hr 1 dataset ~ 4 min @ 0.5 s & 40 kHz resolution > 300,000 images Requirement Part I: Observation database management system Identifying science relevant, best suited dataset Part II: Automated Imaging & calibration pipeline Science ready data products MWA Observation Records Organizer & Navigator (MORON) Querries MWA obs. metadata server Tabulates imp info Tabulates parallel events & conditions of day Cross matches MORON Thesis datasets Imaging AIRCARS Automated SS imaging HDR > 1000; Best = Catalogues Mondal, S. , Mohan, A., Oberoi D. et al. , 2019 Calibration Calibration Problem Attenuator not well characterised abs. flux calibration issue B is the image boundary contour F is attenuator response. t independent for 4 min data Estimation of F: Full disk imgs at diff epochs of obs. Find F( ) for each. Find mean How to define B ? Finding optimal B Using n boundary definition. ? Procedure: Choose images across band & time Define , Find Vs n in Below n ~ 5, falls ~ 1% in linear part for ~1 Chose n = 6 n = 5 4 6 8 n Error Estimate in Random error < 4 % Systematic error when, Uncertainity in ~ 10% 4 % ( Oberoi, D. et al, 2017 ) 4D datacube & SPREDS 4 D Datacube 4 min dataset @ 160 kHz, 0.5 s ~ 23,000 images. Flux density, 4 D datacube Drastic variation in Top: 10, 30, 50, 70, 90 % of peak Bottom: 3, 5, 10, 30, 50, 70, 90 % of peak SPREDS (SPatially REsolved Dynamic Spectrum) A 2 D slice of 4 D datacube Complete local emission evolution Caveats Flux estimation errors Contamination by bright sources Applications on data Part I : A weak active region jet Part II: A weak microflare Part II: Scientific application Part I: A weak jet Introduction Weak active region jets are quite frequent. Faint thermal signatures unreported in HE event lists. Radio emission: supra-thermal e- beams, moving along open field lines. 2 stream instability Coherent radio emission @ High (We use Zucca et al., 2014 density model) Obs. freq density Height e- beams associated with weak jets are tracers of quiet coronal field dynamics. Ginzburg, Zhelenznyakov, 1958 ; Saint-Hilaire et al. 2013; Reid & Ratcliffe, 2014; Melrose 1970a&b, 2017 Caveat: Propagation effects. Modify all source properties No efficient way to disentangle intrinsic dynamics from propagation effects The Event 3 Nov, 2014 (06:12:00 – 06:16:00 UT) - Day of medium activity Observation band: 15 MHz centred at 118 MHz. 6 Groups of type III bursts. (No report in RHESSI or GOES ) AIA 94 MWA Jet Dynamic maps Time res: 0.5 sec ; freq res: 160 kHz ( AIRCARS) 06:12:00 – 06:16:00 UT & 15 MHz about 118 MHz (~ 23,000 maps) Burst sources well described by 2D Gaussian function. 118 MHz Studied Evolution of Gaussian parameters in t & Area, Source sky location, position angle, Integrated flux density & Peak flux density. Mohan, A. et al., ApJ, 2019. 0.3%, 0.5%, 1%, 3%, 5%, 10%, 30%,60% & 90% During the burst We present burst groups: Group I: 06:13:01 - 06:13:20 Group II: 06:14:08 - 06:14:25 Similar The

PhD Defense

Transcript: Hitziger et al. Electro-metabolic coupling investigated with jitter invariant dictionary learning, HBM 2014. Results SNR [dB] presented by Sebastian Hitziger Bruno Torrésani Alain Rakotomamonjy template vs. random Coefficient updates = sparse coding Matching pursuit (MP) Epoching (often manually): problematic if Response onsets are unknown Responses overlap Averaging: problematic if Response latencies vary Response shape changes alternate Hitziger et al. Jitter-adaptive dictionary learning - application to multi-trial neuroelectric signals, ICLR 2013. Collaborators Compared techniques Contiguous AWL from http://jonlieffmd.com Invited guest Main contributions Single long signal Multiple occurrences (overlaps) Variable latencies and durations Implementation: MP (detection) MC-Spike/AD-Spike Lasso problem maximal spiking potential Statistical independence of components Combine previous models Evaluation Stepwise activation using LARS Every step: ensure uniqueness (and non-negativity) AD-Spike/MC-Spike benefit from relearning bad template Difficult to know a priori Hierarchical approach Task-dependent stopping criterion Large noise levels (background activity) No ground truth -> Adequate modeling -> Statistical learning from data Epoching + averaging too simplistic Signal variability (amplitudes, latencies, wave shapes) Complex data (multi-channel, multi-trial, multi-modal) CBF activity around 1 Hz (respiration) Local spiking rates (LSR) match CBF activity Spikes synchronize, phase-locked to CBF rhythm Examiners Short epochs/trials Uniqueness constraint Jitter compensation Implementation: LARS General model Generic algorithm Coefficient updates High spiking activity = high CBF level MC-Spike Detection step Reviewers time to previous spike [s] Use matching pursuit Ensure minimal distance constraint Sandrine Saillet Alexandre Gramfort Christian G. Bénar Bruno Torrésani Tallon-Baudry and Bertrand (1999) Each method addresses only one type of variability Laure Blanc-Féraud Christian G. Bénar Coefficients provide clustering Spike-to-spike distances correlate with spike energies Number of waveforms No epoching Multiple responses Sparsity of components 6) Conclusion Processing full LFP recording Dynamic time warping Speech processing (Itakura, 1975; Sakoe and Chiba, 1971, 1978) Event-related potentials (Picton et al., 1988) Latency compensation Multi-component models Time-frequency representations frequency [Hz] Summation of electromagnetic fields Alternate minimization Woody (1967) Weighted average over isolated spikes Normalization + alignment of peaks Advisors No individual CBF response visible Non-linear summation of responses Overlapping effects: epoching problematic Multi-class spike learning (no dilations) Hierarchical structure, start with alternate Maximizes variance Orthogonality between components Shortcomings No explicit modeling of temporal variability Create dictionary Dictionary learning (DL): http://biomedicalengineering.yolasite.com/neurons.php Hitziger et al. Adaptive waveform learning - application to single- and multi-modal neurological data, in preparation. Adaptive waveform learning (AWL) The general model Jung et al. (2000) 3) Adaptive Waveform Learning (AWL) Using noisy spike template Rescaling of time axis Representations in Fourier or wavelet bases Meaningful representations: time-frequency Complex values: separate phase and amplitude Finite set : discretization of "variability space" Sum over p: multiple occurrences per waveform PhD Defense April 14, 2015 Fast Spiking Rates Coefficients Waveform updates 1) Neural Activity in the Brain 4) Epoched AWL Spikes well isolated Clear CBF response after each spike Modeling the Variability of Electrical Activity in the Brain LFP-CBF recording Within ANR project Multimodel Recording in 6 anesthetized rats with bicuculline injection (epilepsy model) Goal: develop/evaluate models explaining parameter couplings Realigned average for every waveform Normalization + centering Discussion E-AWL specialization Insightful representations Clear artifact separation Comparison to PCA, ICA, template matching Deformations of signal components Coefficient updates Convex problem Solve sequentially for each : generalized averaging Normalization + centering Essentially averaging Choice of right basis Multi-channel extension (preliminary results Papageorgakis, 2014) Different specializations/applications (MEG, detect sleep spindles, ...) More general deformations (e.g., dynamic time warping) Time frequency version Hemodynamic coupling more studies needed Interneuronal communication S. Hitziger, M. Clerc, S. Saillet, A. Gramfort, C. Bénar, T. Papadopoulo. Electro-metabolic coupling investigated with jitter invariant dictionary learning, International Human Brain Mapping Conference (HBM), 2014. S. Hitziger, M. Clerc, S. Saillet, A. Gramfort, C. Bénar, T. Papadopoulo. Jitter-adaptive dictionary learning - application to multi-trial neuroelectric signals, International

PhD defense

Transcript: The problem to be addressed is to perform 3D data registration of an object within a scene using a network of cameras and inertial sensors. Top view Motivations 'Height' indicates the error in the height measuring process (input), shown in cm X, Y and Z are the elements of t vector (output), shown in cm Performing 3D data registration and scene reconstruction using a set of planar images is still one of the key challenges of computer vision. Average processing times in ms for different size of inertial planes. Infinite homography is used to construct image plane of virtual camera. Homography is used to project virtual images onto Euclidean planes in the scene. A person within the scene is being observed by three cameras Extrinsic Parameters Among Cameras A network of cameras, whose usage and ubiquitousness have been increasing in the last decade, can provide such planar images from different views of the scene. x and y indicate the errors in image correspondences (input), shown in pixel X, Y and Z are the elements of t vector (output), shown in cm 3D reconstruction is obtained by stacking several inertial planes Architecture Modelization of Uncertainties Recently, IS has been becoming much cheaper and more available so that nowadays most smart-phones are equipped in both IS and camera sensors. 3D earth cardinal orientation (North-East-Down) is one of the outputs of an IS. Keeping the intersection of shadows The yellow parts are implemented on GPU How can we benefit from having a network of IS and camera couples, for the purpose of 3D data registration? Extension to a network IS' noise is assumed by what is provided by the manufacturer X, Y and Z are the elements of t vector, shown in cm Data of 3rd camera-IS couple 2D images Each point on a virtual image plane conveys an uncertainty due to IS noise Each projected point onto an inertial plane in the scene conveys some uncertainties due to IS noise plus translation vector's noise Having a model of such uncertainties is of importance in fusion stage of the points projected from different cameras and moreover for a further application which will use the final registered points We used statistical geometry to modelized Problem Statement Considering each camera as a light projector, there will appear three shadows on inertial plane 3D orientations 3D reconstruction of the person A set of camera-IS couples can be used to cover the scene from different views PDF (Probability Distribution Function) of results (estimated t) Translation recovery Volumetric reconstruction of a scene normally is a heavy process and time consuming We developed a real-time algorithm using CUDA enabled GP-GPU. Data of 1st camera-IS couple Data of 2nd camera-IS couple It provides the cross-section of the inertial plane with the object. Two arbitrary points in the scene are selected. Their height w.r.p. one camera is measured. The translation t is estimated. Among the extrinsic parameters among two cameras, namely rotation R and translation t, the R is already relaxed once we used the concept of virtual cameras. What remains is t, for which an effective method is introduced to recover. t = f(the height of two arbitrary points w.r.t. one camera, image correspondences) Target applications: Surveillance, human behaviour modelling, virtual-reality, smart-room, health-care, games, teleconferencing, human-robot interaction, medical industries, and scene and object understanding. Average processing times in ms for different number of inertial planes. Illustrative example Real-time implementation using GP-GPU Overall view The certainty of the proposed method has been empirically in three different impacts: Noise of IS 3D observation Heights measurement errors Extraction of image correspondences Geometry behind 3D orientation of IS is fused with 2D image to provide an earth-aligned virtual camera. The orientation is used to virtually generate a set of Euclidean planes in the scene. The 3D data registration is obtained by projection of virtual images onto the Euclidean planes.

Now you can make any subject more engaging and memorable