Catalog start Subject "Overfitting"

%{search_type} search results

6 catalog results

RSS feed for this result
1 - 6
View results as:
Number of results to display per page

Online 1. Moment-Based Probability Modelling and Extreme Response Estimation: The FITS Routine [1998]

Description
Book
Summary
This report documents the usage of the routine FITS, which provides automated fits of various analytical, commonly used probability models from input data. This routine is intended to complement the previously distributed routine, FITTING, documented in RMS Report 14 (Winterstein et al, 1994). The FITTING routine implements relatively complex, four-moment distribution models, whose parameters are fit with numerical optimization routines. While these four-moment fits can be quite useful and faithful to the observed data, their complexity can make them difficult to automate within standard fitting algorithms. In contrast, the routine FITS is intended to provide more robust (lower moment) fits of simpler, more conventional distribution forms. For each database of interest, the routine estimates the distribution of annual maximum response, based on the data values and the duration, T, over which they were recorded. To focus on upper tails of interest, the user can also supply an arbitrary lower-bound threshold,Ιlow , above which a shifted distribution model-exponential or Weibull-is fit. (In estimating the annual maximum response, the program automatically adjusts for the decreasing rate of response events as the threshold Ιlow is raised.) This report generalizes an earlier report (Stanford RMS Report 19; Winterstein , 1995), which introduced the FITS routine. The major extension included in this updated version is the inclusion of a new, "quadratic Weibull'' distribution. This distribution, fitted to the first three moments of a data set, has been found especially useful in modelling fatigue loads observed on wind turbine blades (Lange and Winterstein, 1996). At the same time, unlike the four-moment models from FITTING, the parameter fitting of the quadratic Weibull does not require numerical optimization. It also avoids the tendency of four-moment models toward overfitting, when applied to positive variables such as load peaks (or ranges). This is demonstrated here through an additional example, applying FITS to a data set of observed wind turbine blade loads.
Digital collection
Reliability of Marine Structures Program

2. Second order derivatives for network pruning: Optimal Brain Surgeon

Description
Book
SAL3 (off-campus storage)

3. Quantifying Multivariate Classification Performance - the Problem of Overfitting [electronic resource]. [1999]

  • Washington, D.C. : United States. Dept. of Energy ; Oak Ridge, Tenn. : distributed by the Office of Scientific and Technical Information, U.S. Dept. of Energy, 1999
Description
Book — 1 online resource (11 pages ) : digital, PDF file.
Summary
We have been studying the use of spectral imagery to locate targets in spectrally interfering backgrounds. In making performance estimates for various sensors it has become evident that some calculations are unreliable because of overflying. Hence, we began a thorough study of the problem of overfitting in multivariate classification. In this paper we present some model based results describing the problem. From the model we know the ideal covariance matrix, the ideal discriminant vector, and the ideal classification performance. We then investigate how experimental conditions such as noise, number of bands, and number of samples cause discrepancies from the ideal results. We also suggest ways to discover and alleviate overfitting.
Online

4. Algorithms for reinforcement learning [electronic resource] [2010]

  • Szepesvári, Csaba.
  • San Rafael, Calif. (1537 Fourth Street, San Rafael, CA 94901 USA) : Morgan & Claypool, c2010.
Description
Book — 1 electronic text (xii, 89 p.) : ill.
Summary
  • Markov Decision Processes Value Prediction Problems Control For Further Exploration.
  • (source: Nielsen Book Data)
Reinforcement learning is a learning paradigm concerned with learning to control a system so as to maximize a numerical performance measure that expresses a long-term objective. What distinguishes reinforcement learning from supervised learning is that only partial feedback is given to the learner about the learner's predictions. Further, the predictions may have long term effects through influencing the future state of the controlled system. Thus, time plays a special role. The goal in reinforcement learning is to develop efficient learning algorithms, as well as to understand the algorithms' merits and limitations. Reinforcement learning is of great interest because of the large number of practical applications that it can be used to address, ranging from problems in artificial intelligence to operations research or control engineering. In this book, we focus on those algorithms of reinforcement learning that build on the powerful theory of dynamic programming. We give a fairly comprehensive catalog of learning problems, describe the core ideas, note a large number of state of the art algorithms, followed by the discussion of their theoretical properties and limitations.
(source: Nielsen Book Data)
Online

5. Algorithms for reinforcement learning [2010]

Description
Book — 1 online resource (xii, 89 pages) : illustrations
Summary
  • Markov Decision Processes Value Prediction Problems Control For Further Exploration.
  • (source: Nielsen Book Data)
Reinforcement learning is a learning paradigm concerned with learning to control a system so as to maximize a numerical performance measure that expresses a long-term objective. What distinguishes reinforcement learning from supervised learning is that only partial feedback is given to the learner about the learner's predictions. Further, the predictions may have long term effects through influencing the future state of the controlled system. Thus, time plays a special role. The goal in reinforcement learning is to develop efficient learning algorithms, as well as to understand the algorithms' merits and limitations. Reinforcement learning is of great interest because of the large number of practical applications that it can be used to address, ranging from problems in artificial intelligence to operations research or control engineering. In this book, we focus on those algorithms of reinforcement learning that build on the powerful theory of dynamic programming. We give a fairly comprehensive catalog of learning problems, describe the core ideas, note a large number of state of the art algorithms, followed by the discussion of their theoretical properties and limitations.
(source: Nielsen Book Data)
Online

6. Data Science for Supply Chain Forecasting [2021]

  • Vandeput, Nicolas, author. Author http://id.loc.gov/vocabulary/relators/aut
  • 2nd ed - Berlin ; Boston : De Gruyter, [2021]
Description
Book — 1 online resource (XXVIII, 282 pages) : Digital: text file; PDF.
Summary
  • Frontmatter
  • Acknowledgments
  • About the Author
  • Foreword - Second Edition
  • Foreword - First Edition
  • Contents
  • Introduction
  • Part I: Statistical Forecasting
  • 1 Moving Average
  • 2 Forecast KPI
  • 3 Exponential Smoothing
  • 4 Underfitting
  • 5 Double Exponential Smoothing
  • 6 Model Optimization
  • 7 Double Smoothing with Damped Trend
  • 8 Overfitting
  • 9 Triple Exponential Smoothing
  • 10 Outliers
  • 11 Triple Additive Exponential Smoothing
  • Part II: Machine Learning
  • 12 Machine Learning
  • 13 Tree
  • 14 Parameter Optimization
  • 15 Forest
  • 16 Feature Importance
  • 17 Extremely Randomized Trees
  • 18 Feature Optimization #1
  • 19 Adaptive Boosting
  • 20 Demand Drivers and Leading Indicators
  • 21 Extreme Gradient Boosting
  • 22 Categorical Features
  • 23 Clustering
  • 24 Feature Optimization #2
  • 25 Neural Networks
  • Part III: Data-Driven Forecasting Process Management
  • 26 Judgmental Forecasts
  • 27 Forecast Value Added
  • Now It's Your Turn!
  • A Python
  • Bibliography
  • Glossary
  • Index
Using data science in order to solve a problem requires a scientific mindset more than coding skills. Data Science for Supply Chain Forecasting, Second Edition contends that a true scientific method which includes experimentation, observation, and constant questioning must be applied to supply chains to achieve excellence in demand forecasting. This second edition adds more than 45 percent extra content with four new chapters including an introduction to neural networks and the forecast value added framework. Part I focuses on statistical "traditional" models, Part II, on machine learning, and the all-new Part III discusses demand forecasting process management. The various chapters focus on both forecast models and new concepts such as metrics, underfitting, overfitting, outliers, feature optimization, and external demand drivers. The book is replete with do-it-yourself sections with implementations provided in Python (and Excel for the statistical models) to show the readers how to apply these models themselves. This hands-on book, covering the entire range of forecasting-from the basics all the way to leading-edge models-will benefit supply chain practitioners, forecasters, and analysts looking to go the extra mile with demand forecasting. Check out the webinar on the book, which discusses the general issues and challenges of demand forecasting. The panelists have extensive professional experience in this area, providing insights into best practices (process, models) and discussing how data science and machine learning impact those forecasts: https://www.youtube.com/watch?v=1rXjXcabW2s&t=2s.
(source: Nielsen Book Data)
Online

Articles+

Journal articles, e-books, & other e-resources
See article+ results

Guides

Course- and topic-based guides to collections, tools, and services.
Search ""Overfitting"" in all guide pages