Practical Considerations

Slides:



Advertisements
Similar presentations
Classification Classification Examples
Advertisements

Imbalanced data David Kauchak CS 451 – Fall 2013.
CHAPTER 21 Inferential Statistical Analysis. Understanding probability The idea of probability is central to inferential statistics. It means the chance.
Logistic Regression Rong Jin. Logistic Regression Model  In Gaussian generative model:  Generalize the ratio to a linear model Parameters: w and c.
Overview of The Operations Research Modeling Approach.
Document Classification Comparison Evangel Sarwar, Josh Woolever, Rebecca Zimmerman.
Chapter Sampling Distributions and Hypothesis Testing.
C82MCP Diploma Statistics School of Psychology University of Nottingham 1 Overview of Lecture Independent and Dependent Variables Between and Within Designs.
PROBABILITY AND SAMPLES: THE DISTRIBUTION OF SAMPLE MEANS.
Statistics 03 Hypothesis Testing ( 假设检验 ). When we have two sets of data and we want to know whether there is any statistically significant difference.
Chapter 7 Probability and Samples: The Distribution of Sample Means
DQOs and the Development of MQOs Carl V. Gogolak USDOE Environmental Measurements Lab.
Introduction to the design (and analysis) of experiments James M. Curran Department of Statistics, University of Auckland
INFERENTIAL STATISTICS – Samples are only estimates of the population – Sample statistics will be slightly off from the true values of its population’s.
CSCI 347 / CS 4206: Data Mining Module 06: Evaluation Topic 01: Training, Testing, and Tuning Datasets.
Basic Data Mining Techniques
Hypothesis Testing.
Data Mining Practical Machine Learning Tools and Techniques Slides for Chapter 5 of Data Mining by I. H. Witten, E. Frank and M. A. Hall 報告人:黃子齊
ENSEMBLE LEARNING David Kauchak CS451 – Fall 2013.
Being a Health Literate Consumer (3:04) Click here to launch video Click here to download print activity.
Hyperparameter Estimation for Speech Recognition Based on Variational Bayesian Approach Kei Hashimoto, Heiga Zen, Yoshihiko Nankaku, Akinobu Lee and Keiichi.
CS 782 – Machine Learning Lecture 4 Linear Models for Classification  Probabilistic generative models  Probabilistic discriminative models.
Tahir Mahmood Lecturer Department of Statistics. Outlines: E xplain the role of sampling in the research process D istinguish between probability and.
The delta rule. Learn from your mistakes If it ain’t broke, don’t fix it.
Jen-Tzung Chien, Meng-Sung Wu Minimum Rank Error Language Modeling.
Chapter 7: The Distribution of Sample Means. Frequency of Scores Scores Frequency.
T tests comparing two means t tests comparing two means.
Data Mining Practical Machine Learning Tools and Techniques By I. H. Witten, E. Frank and M. A. Hall Chapter 5: Credibility: Evaluating What’s Been Learned.
Building Valid, Credible & Appropriately Detailed Simulation Models
RESEARCH METHODS Lecture 28. TYPES OF PROBABILITY SAMPLING Requires more work than nonrandom sampling. Researcher must identify sampling elements. Necessary.
COMPLIMENTARY TEACHING MATERIALS Farm Business Management: The Fundamentals of Good Practice Peter L. Nuthall.
Research Design
15 Inferential Statistics.
Sampling and Sampling Distribution
Data Science Credibility: Evaluating What’s Been Learned
Chapter 7. Classification and Prediction
Deep Learning Amin Sobhani.
Randomness in Neural Networks
RESEARCH METHODS Lecture 28
Chapter 4 Marketing Research
MIRA, SVM, k-NN Lirong Xia. MIRA, SVM, k-NN Lirong Xia.
Chapter 4 Marketing Research
Classification: Logistic Regression
Al-Imam Mohammad Ibn Saud University Large-Sample Estimation Theory
Perceptrons Lirong Xia.
LECTURE NOTE By DR. O. J. AKINYOMI
Hypothesis Testing: Hypotheses
Random walk initialization for training very deep feedforward networks
CS 4/527: Artificial Intelligence
REMOTE SENSING Multispectral Image Classification
Pattern Classification All materials in these slides were taken from Pattern Classification (2nd ed) by R. O. Duda, P. E. Hart and D. G. Stork, John.
Discrete Event Simulation - 4
Uncertainty of Measurement
Chapter 11 Practical Methodology
UNDERSTANDING RESEARCH RESULTS: STATISTICAL INFERENCE
Tuning CNN: Tips & Tricks
Business Intelligence
Authors: Wai Lam and Kon Fan Low Announcer: Kyu-Baek Hwang
LECTURE 23: INFORMATION THEORY REVIEW
Chapter 7: The Distribution of Sample Means
Sampling.
Research Problem: The research problem starts with clearly identifying the problem you want to study and considering what possible methods will affect.
MECH 3550 : Simulation & Visualization
Design Of Experiment Eng. Ibrahim Kuhail.
Where does the error come from?
Determining the Size of a Sample
Introduction to the design (and analysis) of experiments
MIRA, SVM, k-NN Lirong Xia. MIRA, SVM, k-NN Lirong Xia.
Variables, Constants, Assign.
Perceptrons Lirong Xia.
Presentation transcript:

Practical Considerations C. Lee Giles Thanks to Alexander Ororbia and Sargur Srihari

Limits to performance Bayes error defines the minimum performance error you can obtain even for infinite training data and know the true probability distribution You input features may not contain complete information about the output variable or your system is intrinsically stochastic You will have a limited amount of training data Collecting data can be expensive

Metric choices Appropriate for the problem Usually from the cost function used to train the model Accuracy or error rate of the system Advanced metrics may be needed

Measuring Performance Reasonable level of performance? Often problem dependent Academic problem Error rates based on previous benchmark data set Or totally new problem with new data set Real-world setting Error rate such that the application is safe, cost-effective, appeals to customers, etc. Design decisions follow Is it more costly to make one kind of mistake than another Spam classification

A region of interest (ROI) are samples within a data set identified for a particular purpose.

or Delta-RNN)

Conclusions/What Did We Learn? Know your performance metrics Don’t forget your baseline models (for proper comparison) Gathering more data can be important Hyperparameter selection is important (i.e., tuning) Debug, debug, debug!!

Questions? Deep robots! Deep questions?!