1,721,028 research outputs found
Chain rules for Rademacher complexity
Get PDFTwo preliminary estimates are obtained for expected suprema of Bernoulli processes indexed by an image of a bounded Euclidean subset through a class of Lipschitz functions. If that bounded subset is given by the projection of a bounded function class onto a sample vector, the result can be considered as a control over empirical Rademacher complexity of a composite function class. Such an estimate can be applied to learning problems where the hypotheses have composite structure or where more than one function is needed to determine the empirical loss.Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2023-08-01The student, Yifeng Chu, accepted the attached license on 2021-07-22 at 15:37.The student, Yifeng Chu, submitted this Thesis for approval on 2021-07-22 at 15:44.This Thesis was approved for publication on 2021-07-23 at 10:29.DSpace SAF Submission Ingestion Package generated from Vireo submission #17068 on 2022-01-12 at 12:55:36Made available in DSpace on 2022-01-12T22:35:21Z (GMT). No. of bitstreams: 2
CHU-THESIS-2021.pdf: 279267 bytes, checksum: 90c9035baf2e64dab3c72960fe6eb04b (MD5)
LICENSE.txt: 4207 bytes, checksum: 36c9b2bccae3ad130e7e3e412e5324e3 (MD5)
Previous issue date: 2021-07-23Embargo set by: Seth Robbins for item 121152
Lift date: 2024-01-12T22:35:30Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Onl
Rational inattention in control of Markov chains
Get PDFThis thesis poses a general model for optimal control subject to information
constraint, motivated in part by recent work on information-constrained
decision-making by economic agents.
In the average-cost optimal control framework, the general model introduced
in this paper reduces to a variant of the linear-programming representation
of the average-cost optimal control problem, subject to an additional
mutual information constraint on the randomized stationary policy. The resulting
in nite-dimensional convex program admits a decomposition based
on the Bellman error, which is the subject of study in approximate dynamic
programming.
Later, we apply the general theory to an information-constrained variant
of the scalar Linear-Quadratic-Gaussian (LQG) control problem. We give
an upper bound on the optimal steady-state value of the quadratic performance
objective and present explicit constructions of controllers that achieve
this bound. We show that the obvious certainty-equivalent control policy is
suboptimal when the information constraints are very severe, and propose
another policy that performs better in this low-information regime. In the
two extreme cases of no information (open-loop) and perfect information,
these two policies coincide with the optimum.Item withdrawn by Laura Spradlin ([email protected]) on 2014-08-04T14:14:05Z
Item was in collections:
University of Illinois Theses & Dissertations (ID: 1)
No. of bitstreams: 2
Shafieepoorfard_Ehsan.tex: 6259 bytes, checksum: d627d06e872810c2831117945ae574f1 (MD5)
Shafieepoorfard_Ehsan.pdf: 453798 bytes, checksum: dedd3848ce01ddaa50b24066a24e4b50 (MD5)Made available in DSpace on 2015-01-21T19:55:22Z (GMT). No. of bitstreams: 2
Ehsan_Shafieepoorfard.pdf: 453798 bytes, checksum: dedd3848ce01ddaa50b24066a24e4b50 (MD5)
Shafieepoorfard_Ehsan.tex: 6259 bytes, checksum: d627d06e872810c2831117945ae574f1 (MD5)Embargo set by: Seth Robbins for item 73174
Lift date: 2017-01-21T19:56:18Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Only Restriction Lifted for Item 73174 on 2017-01-22T10:15:25Z
A channel emulation viewpoint of coding theorems
Get PDFThis thesis introduces a framework of channel emulation. An emulator is defined as a pair of channels, that converts one channel to another channel with possibly different input and output alphabets. With the concept of channel comparison, we define different types of channel emulation and derive the relationship between these types of emulation. We prove the error probability bound of concatenated codes from the channel emulation viewpoint, and derive the evolution of deficiencies of polar codes. We show that the analysis from the emulation and deficiency viewpoint matches the existing results in coding theory, and expect that this viewpoint can inspire the study of constructing long network codes using a layered black box architecture.Item withdrawn by Laura Spradlin ([email protected]) on 2014-12-10T20:21:12Z
Item was in collections:
University of Illinois Theses & Dissertations (ID: 1)
No. of bitstreams: 1
Guang_Xiaoyu.pdf: 337815 bytes, checksum: 9f4595302a825360a20fb12d387c7e06 (MD5)Made available in DSpace on 2015-01-21T19:56:11Z (GMT). No. of bitstreams: 1
Xiaoyu_Guang.pdf: 337815 bytes, checksum: 9f4595302a825360a20fb12d387c7e06 (MD5)Embargo set by: Seth Robbins for item 73215
Lift date: 2017-01-21T19:56:18Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Only Restriction Lifted for Item 73215 on 2017-01-22T10:15:37Z
Analysis of Bayesian control law for adaptive control
No full textThis thesis analyzes the Bayesian control law for adaptive control proposed
by Ortega and Braun. The problem of concern is as follows: Assume the
agent is put into an unknown environment that is sampled from certain distribution.
If the agent is given priors of environment distribution, dynamics
of each environment, and optimal controller for various environments, then
under Bayesian control law, the agent's behavior would be as optimal as
the true controller tailored to that unknown environment in an asymptotic
sense. Furthermore, the results in the original paper are rewritten in a more
understandable way and several ambiguities are clarifed.U of I Only Restriction set for Item 106004 on 2018-05-22T21:10:14Z with date by [email protected] by Janice Progen ([email protected]) on 2018-05-22T21:17:38Z
No. of bitstreams: 1
ECE499-Sp2018-chu-Yifeng.pdf: 260450 bytes, checksum: 40780069866d3b597380ea85101a9e93 (MD5)Approved for entry into archive by James Hutchinson ([email protected]) on 2018-05-22T21:19:57Z (GMT) No. of bitstreams: 1
ECE499-Sp2018-chu-Yifeng.pdf: 260450 bytes, checksum: 40780069866d3b597380ea85101a9e93 (MD5)Made available in DSpace on 2018-05-22T21:19:57Z (GMT). No. of bitstreams: 1
ECE499-Sp2018-chu-Yifeng.pdf: 260450 bytes, checksum: 40780069866d3b597380ea85101a9e93 (MD5)
Previous issue date: 2018-05Embargo set by: James Hutchinson for item 106004
Lift date: 10000-01-01
Reason: Undergraduate senior research not recommended for open accessUndergraduate senior research not recommended for open accessU of I Onl
Adaptive control of a linear system with quantized state observations
Get PDFThe thesis addresses a problem in networked control systems where quantization is a communication constraint. Control design together with parameter estimation algorithms lead to adaptive control techniques. A first order system with unknown parameters is estimated via projection algorithm recursively. Furthermore, a one-step-ahead control law is applied to regulate the output. In the second part of the thesis, the system with quantization is simplified to a system with white noise disturbance by applying a dithering method. It is shown that system with quantization error can be controlled with the same one-step-ahead control law and projection algorithm.Item withdrawn by Laura Spradlin ([email protected]) on 2014-04-16T14:00:18Z
Item was in collections:
University of Illinois Theses & Dissertations (ID: 1)
No. of bitstreams: 2
1_Kaliyeva_Aray.tex: 63200 bytes, checksum: 10c0cca3f2f1b10c28216ac6fd50e9fc (MD5)
Kaliyeva_Aray.pdf: 420337 bytes, checksum: 8abef2461e2b6ab97d621937ec662651 (MD5)Made available in DSpace on 2014-05-30T16:46:44Z (GMT). No. of bitstreams: 3
Aray_Kaliyeva.pdf: 420337 bytes, checksum: 8abef2461e2b6ab97d621937ec662651 (MD5)
1_Kaliyeva_Aray.tex: 63200 bytes, checksum: 10c0cca3f2f1b10c28216ac6fd50e9fc (MD5)
license.txt: 4063 bytes, checksum: 9b1826d41f33cf85f5abc28c6181207a (MD5
Universal approximation of input-output maps and dynamical systems by neural network architectures
Get PDFIt is well known that feedforward neural networks can approximate any continuous function supported on a finite-dimensional compact set to arbitrary accuracy. However, many engineering applications require modeling infinite-dimensional functions, such as sequence-to-sequence transformations or input-output characteristics of systems of differential equations. For discrete-time input-output maps having limited long-term memory, we prove universal approximation guarantees for temporal convolutional nets constructed using only a finite number of computation units which hold on an infinite-time horizon. We also provide quantitative estimates for the width and depth of the network sufficient to achieve any fixed error tolerance. Furthemore, we show that discrete-time input-output maps given by state-space realizations satisfying certain stability criteria admit such convolutional net approximations which are accurate on an infinite-time scale. For continuous-time input-output maps induced by dynamical systems that are stable in a similar sense, we prove that continuous-time recurrent neural nets are capable of reproducing the original trajectories to within arbitrarily small error tolerance over an infinite-time horizon. For a subset of these stable systems, we provide quantitative estimates on the number of neurons sufficient to guarantee the desired error bound.Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2020-10-02 without embargo termsThe student, Joshua Hanson, accepted the attached license on 2020-07-13 at 17:55.The student, Joshua Hanson, submitted this Thesis for approval on 2020-07-13 at 18:10.This Thesis was approved for publication on 2020-07-15 at 09:26.DSpace SAF Submission Ingestion Package generated from Vireo submission #15597 on 2020-10-02 at 15:13:16Made available in DSpace on 2020-10-07T20:59:50Z (GMT). No. of bitstreams: 3
HANSON-THESIS-2020.pdf: 419080 bytes, checksum: 725f66d8cbb7b3bcc78d1d0731f00cbb (MD5)
ecethesis.zip: 465344 bytes, checksum: 48ca50b57fa458a7685c5e947470f549 (MD5)
LICENSE.txt: 4210 bytes, checksum: dfb0bce9eb4e52d1757ae4d0d53eb487 (MD5)
Previous issue date: 2020-07-1
Modeling of electrical circuit with recurrent neural networks
Get PDFIn this dissertation, a circuit modeling methodology using recurrent neural networks (RNNs) is developed. The methodology covers model structure selection, data generation, training, and model implementation for circuit simulation. Several different RNN structures are investigated and their capabilities in circuit modeling are compared. The stability of RNN in the context of circuit modeling is defined and methods to guarantee stability for some RNN structures are developed. The modeling methodology is supported by test cases showing the accuracy and efficiency of RNN models.Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2021-05-01The student, Zaichen Chen, accepted the attached license on 2019-01-26 at 10:49.The student, Zaichen Chen, submitted this Dissertation for approval on 2019-01-26 at 10:59.This Dissertation was approved for publication on 2019-01-28 at 11:53.DSpace SAF Submission Ingestion Package generated from Vireo submission #13370 on 2019-08-22 at 15:04:00Made available in DSpace on 2019-08-23T20:28:05Z (GMT). No. of bitstreams: 2
CHEN-DISSERTATION-2019.pdf: 3183380 bytes, checksum: e403037fa4b4a2ab35fe8cfffe8b3067 (MD5)
LICENSE.txt: 4209 bytes, checksum: 43307c59f3dd964421f005507b2e55d0 (MD5)
Previous issue date: 2019-01-28Embargo set by: Seth Robbins for item 112078
Lift date: 2021-08-23T20:28:11Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemEmbargo set by: Seth Robbins for item 112078
Lift date: 2021-08-23T20:29:33Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemEmbargo set by: Seth Robbins for item 112078
Lift date: 2021-08-23T20:36:18Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Only Restriction Lifted for Item 112078 on 2021-08-24T09:15:10Z
MMSE estimation from quantized observations in the nonasymptotic regime
Get PDFThis thesis studies MMSE estimation on the basis of quantized noisy observations.
It presents nonasymptotic bounds on MMSE regret due to quantization
for two settings: (1) estimation of a scalar random variable given
a quantized vector of n conditionally independent observations, and (2) estimation
of a p-dimensional random vector given a quantized vector of n
observations (not necessarily independent) when the full MMSE estimator
has a sub-Gaussian concentration property.Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2016-03-02 without embargo termsThe student, Jaeho Lee, accepted the attached license on 2015-12-04 at 10:53.The student, Jaeho Lee, submitted this Thesis for approval on 2015-12-04 at 10:58.This Thesis was approved for publication on 2015-12-04 at 14:15.DSpace SAF Submission Ingestion Package generated from Vireo submission #8933 on 2016-03-02 at 12:51:26Made available in DSpace on 2016-03-02T19:34:20Z (GMT). No. of bitstreams: 2
LEE-THESIS-2015.pdf: 319936 bytes, checksum: 698a60f1b5afdf4dbcdec8ef4f877a56 (MD5)
LICENSE.txt: 4206 bytes, checksum: d0ec992081dd6c5a77c79a9e7654698b (MD5)
Previous issue date: 2015-12-0
Input-to-state stable continuous time recurrent neural networks for transient circuit simulation
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2022-04-06 without embargo termsThe student, Alan Yang, accepted the attached license on 2021-12-10 at 10:44.The student, Alan Yang, submitted this Thesis for approval on 2021-12-10 at 10:53.This Thesis was approved for publication on 2021-12-10 at 12:03.DSpace SAF Submission Ingestion Package generated from Vireo submission #17446 on 2022-04-06 at 17:11:18Made available in DSpace on 2022-04-29T21:35:55Z (GMT). No. of bitstreams: 2
YANG-THESIS-2021.pdf: 3432861 bytes, checksum: 88ffb72a7b7ef4d91d8426c8ba7ed4e4 (MD5)
LICENSE.txt: 4206 bytes, checksum: 1f8aa6221805e4e57a7fede2eca8708d (MD5)
Previous issue date: 2021-12-10This thesis proposes a learning approach for continuous-time recurrent neural network (CTRNN) architectures with zero or one hidden layers that guarantees input-to-state stability (ISS). We propose a model parametrization that guarantees the ISS property with respect to a Lur'e-type ISS Lyapunov function that is learned in conjunction with the model parameters. Our stability constraints impose a physical prior on the learned model, and in some cases improve the convergence of model training. The proposed CTRNN models are used to learn fast-to-simulate transient behavioral models for electronic circuits that can be implemented in the Verilog-A analog behavioral modeling language and simulated in commercial circuit simulators. The proposed CTRNNs are used to learn models of a common-source amplifier and a continuous-time linear equalizer that accurately reproduce the original circuits' behavior when interconnected in circuit configurations not encountered during model training
Robustness and generalization guarantees for statistical learning of generative models
Get PDFWe apply tools from the classical statistical learning theory to analyze theoretical properties of modern machine learning problems that are typically phrased in the context of generative models. By combining standard methods based on the theory of empirical processes with ideas from optimal transport and signal recovery, we formally address the generalization and robustness guarantees for the existing and newly suggested algorithms. More specifically, we consider the following three problems: First, we tackle the problem of domain adaptation, where the training data and the test data are drawn from two distributions that are related but not identical. We devise an empirical risk minimization algorithm based on local worst-case risks, and provide generalization and excess risk guarantees of the learned hypothesis, that are robust to drifts in generative models. Second, we consider the learning of coding schemes, where the goal is to minimize the reconstruction risk of the original signal. It turns out that the task can be viewed as approximating the signal-generating distributions by pushforwards of arbitrary distributions via reconstruction maps. We provide learning guarantees based on the notions of optimal transport and classic statistical learning, using reconstruction errors as hypotheses. Third, we propose a framework of assessing representation learning algorithms by evaluating their estimation capabilities of the representation generating the signal. Using polyhedral estimates from the signal recovery literature, we investigate the provably near-optimal guarantees of the topic model.Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2021-05-01The student, Jaeho Lee, accepted the attached license on 2019-01-15 at 12:21.The student, Jaeho Lee, submitted this Dissertation for approval on 2019-01-15 at 12:28.This Dissertation was approved for publication on 2019-01-15 at 14:39.DSpace SAF Submission Ingestion Package generated from Vireo submission #13358 on 2019-08-22 at 15:03:55Made available in DSpace on 2019-08-23T20:28:03Z (GMT). No. of bitstreams: 3
LEE-DISSERTATION-2019.pdf: 483199 bytes, checksum: 9539329342b03f2b23b6b8c1d77e6101 (MD5)
LICENSE.txt: 4206 bytes, checksum: b2f6ea597681f93ebcd70bc7fec874f5 (MD5)
PROQUEST_LICENSE.txt: 4552 bytes, checksum: f3ef11d7d719c01e9ddc6220d8ac9094 (MD5)
Previous issue date: 2019-01-15Embargo set by: Seth Robbins for item 112074
Lift date: 2021-08-23T20:28:11Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemEmbargo set by: Seth Robbins for item 112074
Lift date: 2021-08-23T20:29:33Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemEmbargo set by: Seth Robbins for item 112074
Lift date: 2021-08-23T20:36:18Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Only Restriction Lifted for Item 112074 on 2021-08-24T09:15:10Z
- …
