| 1. |
|
|
| 2. |
- Binggeli, Christian, et al.
(author)
-
Lyman continuum leakage versus quenching with the James Webb Space Telescope : the spectral signatures of quenched star formation activity in reionization-epoch galaxies
- 2018
-
In: Monthly notices of the Royal Astronomical Society. - : Oxford University Press (OUP). - 0035-8711 .- 1365-2966. ; 479:1, s. 368-376
-
Journal article (peer-reviewed)abstract
- In this paper, we study the effects of a recent drop in star formation rate (SFR) on the spectra of epoch of reionization (EoR) galaxies, and the resulting degeneracy with the spectral features produced by extreme Lyman continuum leakage. In order to study these effects in the wavelength range relevant for the upcoming James Webb Space Telescope (JWST), we utilize synthetic spectra of simulated EoR galaxies from cosmological simulations together with synthetic spectra of partially quenched mock galaxies. We find that rapid declines in the SFR of EoR galaxies could seriously affect the applicability of methods that utilize the equivalent width of Balmer lines and the ultraviolet spectral slope to assess the escape fraction of EoR galaxies. In order to determine if the aforementioned degeneracy can be avoided by using the overall shape of the spectrum, we generate mock NIRCam observations and utilize a classification algorithm to identify galaxies that have undergone quenching. We find that while there are problematic cases, JWST/NIRCam or NIRSpec should be able to reliably identify galaxies with redshifts z similar to 7 that have experienced a significant decrease in the SFR (by a factor of 10-100) in the past 50-100 Myr with a success rate greater than or similar to 85 per cent. We also find that uncertainties in the dust-reddening effects on EoR galaxies significantly affect the performance of the results of the classification algorithm. We argue that studies that aim to characterize the dust extinction law most representative in the EoR would be extremely useful.
|
|
| 3. |
- Blomberg, Niclas, 1986-
(author)
-
On Nuclear Norm Minimization in System Identification
- 2016
-
Licentiate thesis (other academic/artistic)abstract
- In system identification we model dynamical systems from measured data. This data-driven approach to modelling is useful since many real-world systems are difficult to model with physical principles. Hence, a need for system identification arises in many applications involving simulation, prediction, and model-based control.Some of the classical approaches to system identification can lead to numerically intractable or ill-posed optimization problems. As an alternative, it has recently been shown beneficial to use so called regularization techniques, which make the ill-posed problems ‘regular’. One type of regularization is to introduce a certain rank constraint. However, this in general still leads to a numerically intractable problem, since the rank function is non-convex. One possibility is then use a convex approximation of rank, which we will do here.The nuclear norm, i.e., the sum of the singular values, is a popular, convex surrogate of the rank function. This results in a heuristic that has been widely used in e.g. signal processing, machine learning, control, and system identification, since its introduction in 2001. The nuclear norm heuristic introduces a regularization parameter which governs the trade-off between model fit and model complexity. The parameter is difficult to tune, and thecurrent thesis revolves around this issue.In this thesis, we first propose a choice of the regularization parameter based on the statistical properties of fictitious validation data. This can be used to avoid computationally costly techniques such as cross-validation, where the problem is solved multiple times to find a suitable parameter value. The proposed choice can also be used as initialization to search methods for minimizing some criterion, e.g. a validation cost, over the parameter domain.Secondly, we study how the estimated system changes as a function of the parameter over its entire domain, which can be interpreted as a sensitivity analysis. For this we suggest an algorithm to compute a so called approximate regularization path with error guarantees, where the regularization path is the optimal solution as a function of the parameter. We are then able to guarantee the model fit, or, alternatively, the nuclear norm of the approximation, to deviate from the optimum by less than a pre-specified tolerance. Furthermore, we bound the l2-norm of the Hankel singular value approximation error, which means that in a certain subset of the parameter domain, we can guarantee the optimal Hankel singular values returned by the nuclear norm heuristic to not change more (in l2-norm) than a bounded, known quantity.Our contributions are demonstrated and evaluated by numerical examples using simulated and benchmark data.
|
|
| 4. |
- Corral-Lopez, Alberto, 1984-, et al.
(author)
-
Evolution of schooling drives changes in neuroanatomy and motion characteristics across predation contexts in guppies
- 2023
-
In: Nature Communications. - : Springer Nature. - 2041-1723. ; 14
-
Journal article (peer-reviewed)abstract
- One of the most spectacular displays of social behavior is the synchronized movements that many animal groups perform to travel, forage and escape from predators. However, elucidating the neural mechanisms underlying the evolution of collective behaviors, as well as their fitness effects, remains challenging. Here, we study collective motion patterns with and without predation threat and predator inspection behavior in guppies experimentally selected for divergence in polarization, an important ecological driver of coordinated movement in fish. We find that groups from artificially selected lines remain more polarized than control groups in the presence of a threat. Neuroanatomical measurements of polarization-selected individuals indicate changes in brain regions previously suggested to be important regulators of perception, fear and attention, and motor response. Additional visual acuity and temporal resolution tests performed in polarization-selected and control individuals indicate that observed differences in predator inspection and schooling behavior should not be attributable to changes in visual perception, but rather are more likely the result of the more efficient relay of sensory input in the brain of polarization-selected fish. Our findings highlight that brain morphology may play a fundamental role in the evolution of coordinated movement and anti-predator behavior.
|
|
| 5. |
|
|
| 6. |
- Dai, Liang, et al.
(author)
-
An online algorithm for controlling a monotone Wiener system
- 2012
-
In: Proceedings of the 2012 24th Chinese Control and Decision Conference (CCDC). - Piscataway, NJ : IEEE. - 9781457720734 - 9781457720727 ; , s. 1585-1590
-
Conference paper (peer-reviewed)abstract
- This paper proposes and studies an online algorithm ('NORTKNAR') for controlling a monotone Wiener system to a given level. Such systems consist of a FIR model, followed by a monotonically in-or decreasing nonlinear static function. We consider phenomena which obey such system up to stochastic perturbations. We study almost sure convergence under weak regularity assumptions. Theoretical results are complemented by empirical results on the control of a PharmacoKinetics-PharmacoDynamic (PK-PD) system regulating concentrations of levuodopa in the bloodstream. Finally, we indicate how those ideas find application to regulating the rate of events in pulsatile systems.
|
|
| 7. |
|
|
| 8. |
- Dai, Liang
(author)
-
On some sparsity related problems and the randomized Kaczmarz algorithm
- 2014
-
Licentiate thesis (other academic/artistic)abstract
- This thesis studies several problems related to recovery and estimation. Specifically, these problems are about sparsity and low-rankness, and the randomized Kaczmarz algorithm. This thesis includes four papers referred to as Paper A, Paper B, Paper C, and Paper D.Paper A considers how to make use of the fact that the solution to an overdetermined system is sparse. This paper presents a three-stage approach to accomplish the task. We show that this strategy, under the assumptions as made in the paper, achieves the oracle property.In Paper B, a Hankel-matrix completion problem arising in system theory is studied. The use of the nuclear norm heuristic for this basic problem is considered. Theoretical justification for the case of a single real pole is given. Results show that for the case of a single real pole, the nuclear norm heuristic succeeds in the matrix completion task. Numerical simulations indicate that this result does not always carry over to the case of two real poles.Paper C discusses a screening approach for improving the computational performance of the Basis Pursuit De-Noising problem. The key ingredient for this work is to make use of an efficient ellipsoid update algorithm. The results of the experiments show that the proposed scheme can improve the overall time complexity for solving the problem.Paper D studies the choice of the probability distribution for implementing the row-projections in the randomized Kaczmarz algorithm. This relates to an open question in the recent literature. The result proves that a probability distribution resulting in a faster convergence of the algorithm can be found by solving a related Semi-Definite Programming optimization problem.
|
|
| 9. |
- Dai, Liang, et al.
(author)
-
On the nuclear norm heuristic for a Hankel matrix completion problem
- 2015
-
In: Automatica. - : Elsevier BV. - 0005-1098 .- 1873-2836. ; 51, s. 268-272
-
Journal article (peer-reviewed)abstract
- This note addresses the question if and why the nuclear norm heuristic can recover an impulse response generated by a stable single-real-pole system, if elements of the upper-triangle of the associated Hankel matrix are given. Since the setting is deterministic, theories based on stochastic assumptions for low-rank matrix recovery do not apply in the considered situation. A 'certificate' which guarantees the success of the matrix completion task is constructed by exploring the structural information of the hidden matrix. Experimental results and discussions regarding the nuclear norm heuristic applied to a more general setting are also given.
|
|
| 10. |
- Dai, Liang, et al.
(author)
-
On the randomized Kaczmarz algorithm
- 2014
-
In: IEEE Signal Processing Letters. - 1070-9908 .- 1558-2361. ; 21:3, s. 330-333
-
Journal article (peer-reviewed)
|
|
| 11. |
|
|
| 12. |
|
|
| 13. |
|
|
| 14. |
- Falck, Tillmann, et al.
(author)
-
Least-Squares Support Vector Machines for the identification of Wiener-Hammerstein systems
- 2012
-
In: Control Engineering Practice. - : Elsevier BV. - 0967-0661 .- 1873-6939. ; 20:11, s. 1165-1174
-
Journal article (peer-reviewed)abstract
- This paper considers the identification of Wiener-Hammerstein systems using Least-Squares Support Vector Machines based models. The power of fully black-box NARX-type models is evaluated and compared with models incorporating information about the structure of the systems. For the NARX models it is shown how to extend the kernel-based estimator to large data sets. For the structured model the emphasis is on preserving the convexity of the estimation problem through a suitable relaxation of the original problem. To develop an empirical understanding of the implications of the different model design choices, all considered models are compared on an artificial system under a number of different experimental conditions. The obtained results are then validated on the Wiener-Hammerstein benchmark data set and the final models are presented. It is illustrated that black-box models are a suitable technique for the identification of Wiener-Hammerstein systems. The incorporation of structural information results in significant improvements in modeling performance.
|
|
| 15. |
- Giri, Sambit K., et al.
(author)
-
Constraining Lyman continuum escape using Machine Learning
- 2018
-
In: Peering towards Cosmic Dawn. - : Cambridge University Press. - 9781107192461 ; , s. 254-258
-
Conference paper (peer-reviewed)abstract
- The James Webb Space Telescope (JWST) will observe the rest-frame ultraviolet/optical spectra of galaxies from the epoch of reionization (EoR) in unprecedented detail. While escaping into the intergalactic medium, hydrogen-ionizing (Lyman continuum; LyC) photons from the galaxies will contribute to the bluer end of the UV slope and make nebular emission lines less prominent. We present a method to constrain leakage of the LyC photons using the spectra of high redshift (z greater than or similar to 6) galaxies. We simulate JWST/NIRSpec observations of galaxies at z = 6-9 by matching the fluxes of galaxies observed in the Frontier Fields observations of galaxy cluster MACS-J0416. Our method predicts the escape fraction f(esc) with a mean absolute error Delta f(esc) approximate to 0.14. The method also predicts the redshifts of the galaxies with an error approximate to 0.0003.
|
|
| 16. |
- Giri, Sambit K., et al.
(author)
-
Identifying reionization-epoch galaxies with extreme levels of Lyman continuum leakage in James Webb Space Telescope surveys
- 2020
-
In: Monthly notices of the Royal Astronomical Society. - : Oxford University Press (OUP). - 0035-8711 .- 1365-2966. ; 491:4, s. 5277-5286
-
Journal article (peer-reviewed)abstract
- The James Webb Space Telescope (JWST) NIRSpec instrument will allow rest-frame ultraviolet/optical spectroscopy of galaxies in the epoch of reionization (EoR). Some galaxies may exhibit significant leakage of hydrogen-ionizing photons into the intergalactic medium, resulting in faint nebular emission lines. We present a machine learning framework for identifying cases of very high hydrogen-ionizing photon escape from galaxies based on the data quality expected from potential NIRSpec observations of EoR galaxies in lensed fields. We train our algorithm on mock samples of JWST/NIRSpec data for galaxies at redshifts z = 6-10. To make the samples more realistic, we combine synthetic galaxy spectra based on cosmological galaxy simulations with observational noise relevant for z greater than or similar to 6 objects of a brightness similar to EoR galaxy candidates uncovered in Frontier Fields observations of galaxy cluster Abell-2744 and MACS-J0416. We find that ionizing escape fractions (f(esc)) of galaxies brighter than m(AB,1500) approximate to 27 mag may be retrieved with mean absolute error Delta f(esc) approximate to 0.09(0.12) for 24 h (1.5 h) JWST/NIRSpec exposures at resolution R = 100. For 24 h exposure time, even fainter galaxies (m(AB,1500) < 28.5 mag) can be processed with Delta f(esc) approximate to 0.14. This framework simultaneously estimates the redshift of these galaxies with a relative error less than 0.03 for both 24 (m(AB,1500) < 28.5 mag) and 1.5 h (m(AB,1500) < 27 mag) exposure times. We also consider scenarios where just a minor fraction of galaxies attain high f(esc) and present the conditions required for detecting a subpopulation of high-f(esc) galaxies within the data set.
|
|
| 17. |
|
|
| 18. |
- Huang, Xiaolin, et al.
(author)
-
Asymmetric nu-tube support vector regression
- 2014
-
In: Computational Statistics & Data Analysis. - : Elsevier BV. - 0167-9473 .- 1872-7352. ; 77, s. 371-382
-
Journal article (peer-reviewed)abstract
- Finding a tube of small width that covers a certain percentage of the training data samples is a robust way to estimate a location: the values of the data samples falling outside the tube have no direct influence on the estimate. The well-known nu-tube Support Vector Regression (nu-SVR) is an effective method for implementing this idea in the context of covariates. However, the nu-SVR considers only one possible location of this tube: it imposes that the amount of data samples above and below the tube are equal. The method is generalized such that those outliers can be divided asymmetrically over both regions. This extension gives an effective way to deal with skewed noise in regression problems. Numerical experiments illustrate the computational efficacy of this extension to the nu-SVR.
|
|
| 19. |
|
|
| 20. |
|
|
| 21. |
- Kotrschal, Alexander, et al.
(author)
-
Brain size does not impact shoaling dynamics in unfamiliar groups of guppies (Poecilia reticulata)
- 2018
-
In: Behavioural Processes. - : Elsevier BV. - 0376-6357 .- 1872-8308. ; 147, s. 13-20
-
Journal article (peer-reviewed)abstract
- Collective movement is achieved when individuals adopt local rules to interact with their neighbours. How the brain processes information about neighbours' positions and movements may affect how individuals interact in groups. As brain size can determine such information processing it should impact collective animal movement. Here we investigate whether brain size affects the structure and organisation of newly forming fish shoals by quantifying the collective movement of guppies (Poecilia reticulata) from large- and small-brained selection lines, with known differences in learning and memory. We used automated tracking software to determine shoaling behaviour of single-sex groups of eight or two fish and found no evidence that brain size affected the speed, group size, or spatial and directional organisation of fish shoals. Our results suggest that brain size does not play an important role in how fish interact with each other in these types of moving groups of unfamiliar individuals. Based on these results, we propose that shoal dynamics are likely to be governed by relatively basic cognitive processes that do not differ in these brain size selected lines of guppies.
|
|
| 22. |
- Kotrschal, Alexander, et al.
(author)
-
Evolution of brain region volumes during artificial selection for relative brain size
- 2017
-
In: Evolution. - : Wiley. - 0014-3820 .- 1558-5646. ; 71:12, s. 2942-2951
-
Journal article (peer-reviewed)abstract
- The vertebrate brain shows an extremely conserved layout across taxa. Still, the relative sizes of separate brain regions vary markedly between species. One interesting pattern is that larger brains seem associated with increased relative sizes only of certain brain regions, for instance telencephalon and cerebellum. Till now, the evolutionary association between separate brain regions and overall brain size is based on comparative evidence and remains experimentally untested. Here, we test the evolutionary response of brain regions to directional selection on brain size in guppies (Poecilia reticulata) selected for large and small relative brain size. In these animals, artificial selection led to a fast response in relative brain size, while body size remained unchanged. We use microcomputer tomography to investigate how the volumes of 11 main brain regions respond to selection for larger versus smaller brains. We found no differences in relative brain region volumes between large- and small-brained animals and only minor sex-specific variation. Also, selection did not change allometric scaling between brain and brain region sizes. Our results suggest that brain regions respond similarly to strong directional selection on relative brain size, which indicates that brain anatomy variation in contemporary species most likely stem from direct selection on key regions.
|
|
| 23. |
- Kotrschal, Alexander, et al.
(author)
-
Rapid evolution of coordinated and collective movement in response to artificial selection
- 2020
-
In: Science Advances. - : AMER ASSOC ADVANCEMENT SCIENCE. - 2375-2548. ; 6:49
-
Journal article (peer-reviewed)abstract
- Collective motion occurs when individuals use social interaction rules to respond to the movements and positions of their neighbors. How readily these social decisions are shaped by selection remains unknown. Through artificial selection on fish (guppies, Poecilia reticulata) for increased group polarization, we demonstrate rapid evolution in how individuals use social interaction rules. Within only three generations, groups of polarization-selected females showed a 15% increase in polarization, coupled with increased cohesiveness, compared to fish from control lines. Although lines did not differ in their physical swimming ability or exploratory behavior, polarization-selected fish adopted faster speeds, particularly in social contexts, and showed stronger alignment and attraction responses to multiple neighbors. Our results reveal the social interaction rules that change when collective behavior evolves.
|
|
| 24. |
- Liu, Yang, et al.
(author)
-
APTER : Aggregated Prognosis Through Exponential Re-weighting
- 2019
-
In: Computing and Combinatorics. - Cham : Springer Nature. - 9783030261764 - 9783030261757 ; , s. 425-436
-
Conference paper (peer-reviewed)abstract
- This paper considers the task of learning how to make a prognosis of a patient based on his/her micro-array expression levels. The method is an application of the aggregation method as recently proposed in the literature on theoretical machine learning, and excels in its computational convenience and capability to deal with high-dimensional data. This paper gives a formal analysis of the method, yielding rates of convergence similar to what traditional techniques obtain, while it is shown to cope well with an exponentially large set of features. Those results are supported by numerical simulations on a range of publicly available survival-micro-array data sets. It is empirically found that the proposed technique combined with a recently proposed pre-processing technique gives excellent performances. All used software files and data sets are available on the authors' website http://user.it.uu.se/similar to liuya610/index.html
|
|
| 25. |
|
|
| 26. |
- Nygren, Johannes, et al.
(author)
-
A cooperative decentralized PI control strategy : discrete-time analysis and nonlinear feedback
- 2012
-
In: Proc. 3rd IFAC Workshop on Distributed Estimation and Control in Networked Systems. ; , s. 103-108
-
Conference paper (peer-reviewed)abstract
- This paper discusses an extension of a PI control strategy towards the control of a large m×m-MIMO system. This strategy is fully decentralized, it requires only the tuning of m different controllers, while we only allow for neighboring controllers to exchange error signals. This makes it a strong candidate for an implementation on a decentralized, low-power and high performance Wireless Sensor Network (WSN). The main idea is to feed locally observed control errors (’feedback’) not only into the local control law, but also in a fixed proportionate way into neighboring controllers. The analysis concerns convergence to a set point. The analysis is essentially based on a conversion of the PI control law into a discrete-time gradient descent scheme. As an interesting byproduct, this analysis indicates how to deal with quantization functions and nonlinear effects in the feedback signals.
|
|
| 27. |
- Nygren, Johannes, et al.
(author)
-
A direct proof of the discrete-time multivariate circle and Tsypkin criteria
- 2016
-
In: IEEE Transactions on Automatic Control. - 0018-9286 .- 1558-2523. ; 61:2, s. 544-549
-
Journal article (peer-reviewed)abstract
- This technical note presents a new proof of the circle criterion for multivariate, discrete-time systems with time-varying feedback nonlinearities. A new proof for the multivariate Tsypkin criterion for time-invariant monotonic feedback nonlinearities is derived as well. Both integrator- and non-integrator systems are considered. The proofs are direct in the sense that they do not resort to any existing result in systems theory, such as Lyapunov theory, passivity theory or the small-gain theorem. Instead, the proofs refer to the asymptotic properties of block-Toeplitz matrices. One major advantage of the new proof is that it elegantly handles integrator systems without resorting to loop transformation/pole shifting techniques. Additionally, less conservative stability bounds are derived by making stronger assumptions on the sector bound conditions on the feedback nonlinearities. In particular, it is exemplified how this technique relaxes stability conditions of (i) a model predictive control (MPC) rule and (ii) an integrator system.
|
|
| 28. |
- Nygren, Johannes, et al.
(author)
-
A stability criterion for switching Lur'e systems with switching-path restrictions
- 2018
-
In: Automatica. - : Elsevier BV. - 0005-1098 .- 1873-2836. ; 96, s. 337-341
-
Journal article (peer-reviewed)abstract
- This paper derives a stability criterion for switching Lur'e MIMO systems with time-varying nonlinearities, where switching is subject to given restrictions. These restrictions are encoded via the notion of a switching path, interpolating the criterion between the single-system case and arbitrary switching. The criterion for switching Lur'e systems is shown to boil down to a familiar LMI criterion for linear switching systems for the case the Lur'e systems have a linear feedback. The model of a switching Lur'e system is used to describe an MPC control scheme subject to loss of output readings. This example illustrates the practical and theoretical usefulness of the introduced switching path restrictions.
|
|
| 29. |
|
|
| 30. |
- Nygren, Johannes, et al.
(author)
-
Conditions for input-output stability of discrete-time Luré systems with time-varying delays
- 2015
-
In: Proc. 54th Conference on Decision and Control. - Piscataway, NJ : IEEE. - 9781479978861 ; , s. 7707-7714
-
Conference paper (peer-reviewed)abstract
- This paper derives a stability condition for a type of Lur'e systems with time-varying delays and a feedback nonlinearity. The case of discrete-time systems is considered, consisting of a LTI, fed back though a time-varying static nonlinearity. There is an additional delay before or after this nonlinearity, which delays the signals by a positive, time-varying number of steps. Either the time-delay needs to be bounded by a constant (if the LTI system contains a single integrator) or its rate need to be bounded (in case the LTI system is stable). It turns out that, if the LTI has a single integrator with non-decreasing impulse response, the derived stability criterion coincides exactly with the circle criterion for the corresponding constant delay system. The technical proofs rely on direct manipulation of the involved signals, and do not make use of traditional tools as the small gain theorem, Lyapunov functions or passivity results.
|
|
| 31. |
|
|
| 32. |
|
|
| 33. |
- Nygren, Johannes, 1984-
(author)
-
Input-Output Stability Analysis of Networked Control Systems
- 2016
-
Doctoral thesis (other academic/artistic)abstract
- The main focus of the thesis is to derive stability criteria for networked control system (NCS) models featuring imperfections such as time-varying and constant delays, quantization, packet dropouts, and non-uniform sampling intervals. The main method of proof is based on matrix algebra, as opposed to methods using Lyapunov functions or integral quadratic constraints (IQC). This work puts a particular focus on handling systems with a single integrator. This framework is elaborated in different specific directions as motivated by practical realizations of NCSs, as well as through numerical examples. A novel proof of the discrete time multivariate circle criterion and the Tsypkin criterion for systems including a single integrator is presented, as well as a stability criterion for linear systems with a single integrator subject to variable sampling periods and sector-bounded nonlinear feedback. Four stability criteria for different classes of systems subject to packet loss and time-varying delay are given. Stability criteria for a closed loop system switching between a set of linear time-invariant systems (LTIs) are proved. This result is applied to a single-link NCS with feedback subject to packet loss. Finally, necessary and sufficient conditions for delay-independent stability of an LTI system subject to nonlinear feedback are derived.
|
|
| 34. |
- Nygren, Johannes, et al.
(author)
-
Iterative Learning Control and Recursive Identification
- 2010
-
Conference paper (other academic/artistic)abstract
- This abstract discusses our investigations relating Iterative Learning Control (ILC) for periodic systems on the one hand, and the class of Recursive Identification (RI), Gradient Descent (GD), Stochastic Approximation (SA) and adaptive filtering algorithms on the other. The benefit of such is the straightforward transfer of results in the latter context which is useful to study different design decisions made for the former. We discuss briefly the possible relevance of this observation for (i) design and analysis of suitable gain factors, (ii) working with constrained control signals, (iii) designing a model-free control strategy. For a survey of design, analysis and applications of ILC, see e.g. (1; 2). For an overview of practical and theoretical studies of RI and SA see (3), GD (4; 5), and adaptive filtering (6; 7). In this note we articulate the basic idea, and discuss further work which may be expected from this.
|
|
| 35. |
|
|
| 36. |
- Nygren, Johannes
(author)
-
Output feedback control : Some methods and applications
- 2014
-
Licentiate thesis (other academic/artistic)abstract
- This thesis studies some output feedback control laws. Particularly, iterative learning control (ILC) and decentralized network based algorithms are studied. Applications to control of wastewater treatment plants are outlined. For a linear, discrete time MIMO plant, it is shown that the size of the global controller gain, also referred to as the diffusion matrix, plays an important role in stabilization of a decentralized control system with possibly non-linear output feedback. Based on information from a step response experiment of the open loop system, a controller gain which is sufficient for stability can be found. For the SISO case, frequency response expressions are derived for the choice of this controller gain. The results relate nicely to notions of optimality and the Nyquist stability criterion. Various types of ILC algorithms are analysed and numerically illustrated. In particular, new expressions of the asymptotic control error variance for adjoint based iterative learning control (ILC) are derived. It is proven that the control error variance converges to its minimum if a decreasing learning gain matrix is used for ILC. In a simulation study ILC is applied to control a sequencing batch reactor. It is shown numerically that an adjoint based ILC outperforms inverse based ILC and model-free, proportional ILC. A merge of an activated sludge process simulator and a simulator for a wireless sensor network is described and used for illustrating some control performance. Finally, in a numerical optimization study it is shown that the aeration energy can be decreased if many dissolved oxygen sensors are used for aeration control in a biological reactor for nitrogen removal. This results may support future use of inexpensive wireless sensor networks for control of wastewater treatment plants.
|
|
| 37. |
|
|
| 38. |
|
|
| 39. |
- Pelckmans, Kristiaan
(author)
-
An adaptive compression algorithm in a deterministic world
- 2013
-
In: Algorithmic Probability and Friends. - Berlin, Heidelberg : Springer Berlin/Heidelberg. - 9783642449574 ; , s. 299-305
-
Conference paper (peer-reviewed)abstract
- Assume that we live in a deterministic world, we ask ourselves which place the device of randomness still may have, even in case that there is no philosophical incentive for it. This note argues that improved accuracy may be achieved when modeling the (deterministic) residuals of the best model of a certain complexity as 'random'. In order to make this statement precise, the setting of adaptive compression is considered: (1) accuracy is understood in terms of codelength, and (2) the 'random device' relates to Solomonoff's Algorithmic Probability (ALP) via arithmetic coding. The contribution of this letter is threefold: (a) the proposed adaptive coding scheme possesses interesting behavior in terms of its regret bound, and (b) a mathematical characterization of a deterministic world assumption is given. (c) The previous issues then facilitate the derivation of the Randomness-Complexity (RC) frontier of the given algorithm.
|
|
| 40. |
|
|
| 41. |
|
|
| 42. |
|
|
| 43. |
- Pelckmans, Kristiaan
(author)
-
Monitoring High-Frequency Data Streams in FinTech : FADO Versus K-means
- 2020
-
In: IEEE Intelligent Systems. - 1541-1672 .- 1941-1294. ; 35:2, s. 36-42
-
Journal article (peer-reviewed)abstract
- Modern applications of FinTech are challenged by enormous volumes of financial data. One way to handle these is to adopt a streaming setting where data are only available to the algorithms during a very short time. When a new data point (financial transaction) is generated, it needs to be processed directly, and be forgotten immediately after. Especially, ongoing globalization efforts in FinTech require modern methods of fault detection to be able to work efficiently through more than 10 000 financial transactions per second if they are to be deployed as a first line of defence. This article investigates two algorithms able to perform well in this demanding setting: K-means and FADO. Especially, this article provides supports for the claim that “the use of multiple clusters does not necessarily translate into increased detection performance”. To support this claim, results are reported when operating in a quasi-realistic case study of Anti Money Laundering (AML) detection in real-time payment systems. We focus on two prototypical algorithms: the passive aggressive FADO assuming a single cluster, and the well-known K-means algorithm working with K > 1 clusters. We find-in this case-that the use of K-means with multiple clusters is unfavorable as 1) both tuning for K, as well as the need for additional complexity in the K-means algorithm challenges the computational constraints; 2) K-means introduces necessarily added variability (unreliability) in the results; 3) it requires dimensionality reduction, compromising interpretability of the detections; 4) the prevalence of singleton clusters adds unreliability to the outcome. This makes in the presented case FADO favorable over K-means (with K > 1).
|
|
| 44. |
|
|
| 45. |
|
|
| 46. |
|
|
| 47. |
- Pelckmans, Kristiaan
(author)
-
Randomized gossip algorithms for achieving consensus on the majority vote
- 2013
-
In: Proc. 11th IFAC International Workshop on Adaptation and Learning in Control and Signal Processing.
-
Conference paper (peer-reviewed)abstract
- This paper studies a decentralized, randomized gossip algorithm for computing a majority vote amongst the binary decisions associated to n nodes organized in a fixed, ad-hoc network. It is indicated how this problem can be reduced to computing the global average using a standard, randomized gossip algorithm. Then, we illustrate how the majority vote problem allows one to formulate individual stopping rules deciding when an individual node makes its final verdict. Finally, we will provide an illustration of how well the algorithm and associated stopping rule behaves.
|
|
| 48. |
|
|
| 49. |
- Souza, Abel, PhLic. 1986-, et al.
(author)
-
A HPC Co-Scheduler with Reinforcement Learning
-
Other publication (other academic/artistic)abstract
- High Performance Computing (HPC) datacenters process thousands of diverse applications, supporting many scientific and business endeavours. Although users understand minimum coarse resource job requirements such as amounts of CPUs and memory, internal infrastructural utilization data and system dynamics are often visible only to cluster operators. Besides that, due to increased complexity, heuristically tweaking a batch system is even today a very challenge task. When combined with applications profiling, infrastructural data enables improvements to job scheduling, while creating space to improve Quality-of-Service (QoS) metrics such as queue waiting times and total execution times. Targeting improvements in utilization and throughput, in this paper we evaluate and propose a novel Reinforcement Learning co-scheduler algorithm that combines capacity utilization with application performance profiling. We first profile a running application by assessing its resource utilization and progress by means of a forest of decision trees, enabling our algorithm to understand the application’s resource capacity usage. We then use this information to estimate how much capacity from this ongoing allocation can be allocated for co-scheduling additional applications. Because estimations may go wrong, our algorithm has to learn and evaluate when co-scheduling decisions results in QoS degradation, such as application slowness. To overcome this, we devised a co-scheduling architecture and a handful metric to help minimizing performance degradation, enabling improvements on utilization of up to 25% even when the cluster is experiencing high demands, with 10% average queue makespan reductions when experiencing low loads.Together with the architecture, our algorithm forms the base of an application-aware co-scheduler for improved datacenter utilization and minimal performance degradation.
|
|
| 50. |
- Souza, Abel, PhD, 1986-, et al.
(author)
-
A HPC Co-scheduler with Reinforcement Learning
- 2021
-
In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). - Cham : Springer. - 9783030882235 - 9783030882242 ; , s. 126-148
-
Conference paper (peer-reviewed)abstract
- Although High Performance Computing (HPC) users understand basic resource requirements such as the number of CPUs and memory limits, internal infrastructural utilization data is exclusively leveraged by cluster operators, who use it to configure batch schedulers. This task is challenging and increasingly complex due to ever larger cluster scales and heterogeneity of modern scientific workflows. As a result, HPC systems achieve low utilization with long job completion times (makespans). To tackle these challenges, we propose a co-scheduling algorithm based on an adaptive reinforcement learning algorithm, where application profiling is combined with cluster monitoring. The resulting cluster scheduler matches resource utilization to application performance in a fine-grained manner (i.e., operating system level). As opposed to nominal allocations, we apply decision trees to model applications’ actual resource usage, which are used to estimate how much resource capacity from one allocation can be co-allocated to additional applications. Our algorithm learns from incorrect co-scheduling decisions and adapts from changing environment conditions, and evaluates when such changes cause resource contention that impacts quality of service metrics such as jobs slowdowns. We integrate our algorithm in an HPC resource manager that combines Slurm and Mesos for job scheduling and co-allocation, respectively. Our experimental evaluation performed in a dedicated cluster executing a mix of four real different scientific workflows demonstrates improvements on cluster utilization of up to 51% even in high load scenarios, with 55% average queue makespan reductions under low loads.
|
|
