Search
Search Funnelback University
Did you mean economiespast |u:mlg.eng.cam.ac.uk?
101 -
150 of
370
search results for Economics test |u:mlg.eng.cam.ac.uk
where 35
match all words and 335
match some words.
Results that match 1 of 2 words
-
Dependent Indian Buffet Processes Sinead Williamson Peter Orbanz…
https://mlg.eng.cam.ac.uk/pub/pdf/WilOrbGha10.pdf13 Feb 2023: The data was randomly split into atraining set of 130 countries, and a test set of 14 coun-tries. ... For each test country, one randomly selectedindicator was observed, and the remainder held outfor prediction. -
Geometrically Coupled Monte Carlo Sampling Mark Rowland∗University of …
https://mlg.eng.cam.ac.uk/adrian/NeurIPS18-gcmc.pdf19 Jun 2024: 5.2 Variance-reduced ELBO estimation for deep generative models. In this section, we test GCMC sampling strategies on a deep generative modelling application. ... 10. Arthur Gretton, Karsten M. Borgwardt, Malte J. Rasch, Bernhard Schölkopf, and -
Learning to Control a Low-Cost Manipulator usingData-Efficient…
https://mlg.eng.cam.ac.uk/pub/pdf/DeiRasFox11.pdf13 Feb 2023: Althoughdeposit failure feedback was not available to the learner, thedeposit success is good across 10 test trials and four differenttraining setups. ... Second, Tab. III reports the block-deposit success rates for10 test runs (and four different -
Gibbs sampling (an MCMC method) and relations to EM
https://mlg.eng.cam.ac.uk/zoubin/SALD/week7at.pdf27 Jan 2023: Uniform U(0, θ] distribution, θ > 0. We select at random m n disks, having a common θ for failure We select n of these (at random) and test them until -
Probabilistic Modelling, Machine Learning,and the Information…
https://mlg.eng.cam.ac.uk/zoubin/talks/mit12csail.pdf27 Jan 2023: Some Comparisons. Table 1: Test errors and predictive accuracy (smaller is better) for the GP classifier, the supportvector machine, the informative vector machine, and the sparse pseudo-inputGP classifier. ... Data set GPC SVM IVM SPGPC. name train:test -
LNAI 3944 - Evaluating Predictive Uncertainty Challenge
https://mlg.eng.cam.ac.uk/pub/pdf/QuiRasSinetal06.pdf13 Feb 2023: The participants could then usethem to train their algorithms before submitting the test predictions. ... The test results were made public on December 11. The website remainsopen for submission. -
Approximate inference for the loss-calibrated Bayesian
https://mlg.eng.cam.ac.uk/pub/pdf/LacHusGha11.pdf13 Feb 2023: We also assume. the transductive scenario where we are given a test setS of S points {xs}Ss=1, i.e. ... of the shift between the test and trainingdistributions (columns) and the asymmetry of loss (rows). -
Bayesian Hierarchical Clustering Katherine A. Heller…
https://mlg.eng.cam.ac.uk/zoubin/papers/bhcnew.pdf27 Jan 2023: As we will see, the main difference isthat our algorithm uses a statistical hypothesis test tochoose which clusters to merge. ... Sec-ond our algorithm is derived from Dirichlet processmixtures. Third the hypothesis test at the core ofour algorithm tests -
gppl.dvi
https://mlg.eng.cam.ac.uk/zoubin/papers/icml05chuwei-pl.pdf27 Jan 2023: 2.3. Prediction. Now let us take a test pair (r, s) on which the pref-erence relation is unknown. ... our algorithm. The test results of the two algorithmsare presented in the left graph of Figure 2. -
The EM-EP Algorithm forGaussian Process Classification Hyun-Chul Kim� …
https://mlg.eng.cam.ac.uk/zoubin/papers/ecml03.pdf27 Jan 2023: Fig 2 (a) and Fig 2 (b) shows the training set and the test set. ... Each fold were subsequently used as a test set, while the other 9. -
Warped Gaussian Processes Edward Snelson∗ Carl Edward Rasmussen†…
https://mlg.eng.cam.ac.uk/pub/pdf/SneRasGha04.pdf13 Feb 2023: the following table which shows the range of the targets(tmin, tmax), the number of input dimensions (D), and the size of the training and test sets(Ntrain, Ntest) that we ... We show three measuresof performance over independent test sets: mean absolute -
images/test_user_webdesign.eps
https://mlg.eng.cam.ac.uk/pub/pdf/IwaShaGha13a.pdf13 Feb 2023: The vertical axis is the test likelihood, and the horizontal is the test time period. ... Thevertical axis is the test likelihood, and the horizontal is the test time period. -
Draft version; accepted for NIPS*03 Warped Gaussian Processes Edward…
https://mlg.eng.cam.ac.uk/zoubin/papers/gpwarp.pdf27 Jan 2023: the following table which shows the range of the targets(tmin, tmax), the number of input dimensions (D), and the size of the training and test sets(Ntrain, Ntest) that we ... We show three measuresof performance over independent test sets: mean absolute -
Bayesian Classifier Combination Hyun-Chul Kim Zoubin GhahramaniKorea…
https://mlg.eng.cam.ac.uk/pub/pdf/KimGha12.pdf13 Feb 2023: 6500, 1000, 797 data points were selected from the orig-inal test set as a validation set for DNA data set, Satellitedata set, UCI digit data set, respectively. ... BCCresults are based on comparing the posterior mode ofti for data points in the test set -
A Probabilistic Model for Online Document Clustering with Application …
https://mlg.eng.cam.ac.uk/pub/pdf/ZhaGhaYan04a.pdf13 Feb 2023: In addition to the binary decision “novel” or “non-novel”, eachsystem is required to generated a confidence score for each test document. ... References. [1] The 2002 topic detection & tracking task definition and evaluation -
chu05a.dvi
https://mlg.eng.cam.ac.uk/zoubin/papers/chu05a.pdf27 Jan 2023: 0.23370.0072. Table 2: Test results of the three algorithms using a Gaussian kernel. ... Figure 4 presents the test results of the three algorithms for different numbers ofselected genes. -
Learning to Parse Images
https://mlg.eng.cam.ac.uk/pub/pdf/HinGhaTeh99a.pdf13 Feb 2023: Then the learning. 2,3. 2,4. 2,5. 3,4. 3,5. 4,5. Figure 1: Sample images from the test set. ... tested on the same test set. -
LNAI 8189 - Variational Hidden Conditional Random Fields with Coupled …
https://mlg.eng.cam.ac.uk/pub/pdf/BouZafMor13a.pdf13 Feb 2023: In each case, we evaluated their performance on a test set consist-ing of sequences from 3 debates. ... Since we have introduced parameters θ it is sensible to test our methodologyfor signs of overfitting. -
Nonparametric Transforms of Graph Kernelsfor Semi-Supervised Learning …
https://mlg.eng.cam.ac.uk/zoubin/papers/ZhuKanGhaLaf04.pdf27 Jan 2023: Morespecifically, we restrict ourselves to thetransductive setting where the unlabeled data alsoserve as the test data. ... All classes must be present in the labeled set. The rest is used asunlabeled (test) set in that trial. -
Gaussian Processes in Reinforcement Learning Carl Edward Rasmussen…
https://mlg.eng.cam.ac.uk/pub/pdf/RasKus04.pdf13 Feb 2023: The predictive distribution for a novel test input. is Gaussian: ) 2! ... on random examples, the relations can already be approximated to within root meansquared errors (estimated on -& test samples and considering the mean of the predicteddistribution) -
You Shouldn’t Trust Me: Learning Models WhichConceal Unfairness From…
https://mlg.eng.cam.ac.uk/adrian/ECAI20-You_Shouldn%E2%80%99t_Trust_Me.pdf19 Jun 2024: Each histogramrepresents the ranking across the test set assigned by the designated feature importance method. ... These results suggest that ourattack is successful in generalising across unseen test points. -
The Infinite Hidden Markov Model Matthew J. Beal Zoubin ...
https://mlg.eng.cam.ac.uk/pub/pdf/BeaGhaRas02.pdf13 Feb 2023: We propose estimating the likelihood of a test sequence given a learned model using particlefiltering. ... 6Different particle initialisations apply if we do not assume that the test sequence immediatelyfollows the training sequence. -
Computational abstractions for probabilistic and differentiable…
https://mlg.eng.cam.ac.uk/capp-workshop/23 Nov 2022: Online. Bayesian inference and machine learning have found numerous use cases in applied domains and basic science, such as disease modeling, climate research, economics, or astronomy. -
Ten papers from the group to appear at ICML 2016 | Cambridge Machine…
https://mlg.eng.cam.ac.uk/news/ten-papers-from-the-group-to-appear-at-icml-2016/3 Jul 2024: Train and Test Tightness of LP Relaxations in Structured Prediction. Ofer Meshi, Mehrdad Mahdavi, Adrian Weller and David Sontag. -
Bayesian Deep Learning via Subnetwork Inference · Cambridge MLG Blog
https://mlg.eng.cam.ac.uk/blog/2021/07/21/subnetwork-inference.html12 Apr 2024: Figure 9: Results on the rotated MNIST benchmark, showing the mean $pm$ std of the test error (top) and log-likelihood (bottom) across three different seeds. ... methods. Figure 10: Results on the corrupted CIFAR-10 benchmark, showing the mean $pm$ std -
Speaking Truth to Climate Change
https://mlg.eng.cam.ac.uk/carl/climate/truth.pdf4 Jul 2024: The effect of the alliance is to immediately apply strong economic pressure on allcontries to reduce emissions. ... Alliance dynamics. Initially, from a purely economic perspective, it’ll be advantageous for low percapita emitting countries to join -
Natural-Gradient Variational Inference 2: ImageNet-scale · Cambridge…
https://mlg.eng.cam.ac.uk/blog/2021/11/24/ngvi-bnns-part-2.html12 Apr 2024: Reducing the prior precision $delta$ results in higher validation accuracy, but also a larger train-test gap, corresponding to more overfitting. ... Continual Learning: I personally think continual learning is a very good way to test approximate Bayesian -
4F13 Machine Learning: Coursework #3: Latent Dirichlet Allocation…
https://mlg.eng.cam.ac.uk/teaching/4f13/1819/cw/coursework3.pdf19 Nov 2023: c) For the Bayesian model, what is the log probability for the test document with ID 2001? ... Explainwhether, when computing the log probability of a test document, you would use the multinomial withor without the “combinatorial factor”. -
Modelling data
https://mlg.eng.cam.ac.uk/teaching/4f13/1819/modelling%20data.pdf19 Nov 2023: generalize from observations in the training set to new test cases(interpolation and extrapolation). • ... make predictions on test cases• interpret the trained model, what insights is the model providing?• evaluate the accuracy of model. • -
Infinite Hidden Markov Models and extensions
https://mlg.eng.cam.ac.uk/zoubin/talks/BayesHMMs10.pdf27 Jan 2023: IHMM evaluation in (Beal et al.,2002) is more elaborate: it allows the IHMM to con-tinue learning about new data encountered during test-ing. ... Subjecting theseresults to the same analysis as the artificial data re-veals similar compared test-set -
standalone.dvi
https://mlg.eng.cam.ac.uk/pub/pdf/QuiRasWil07.pdf13 Feb 2023: test conditional: p(f|u) = N (K,uK1. u,uu, K, Q,) , (10b). ... Lower graph: assumption of conditional independencebetween training and test function values given u. -
- 4F13: Machine Learning
https://mlg.eng.cam.ac.uk/teaching/4f13/0809/lect06.pdf19 Nov 2023: Constraint-Based Learning: Use statistical tests of marginal and conditionalindependence. Find the set of DAGs whose d-separation relations match theresults of conditional independence tests. -
Variational Inference for the Indian Buffet Process Finale…
https://mlg.eng.cam.ac.uk/pub/pdf/DosMilVanTeh09.pdf13 Feb 2023: Table 1: Running times in seconds and test log-likelihoods for the Yale Faces dataset. ... Algorithm K Time Test Log-Likelihood. 2 56 -0.7444Finite Gibbs 5 120 -0.4220. -
A Kernel Approach to Tractable Bayesian Nonparametrics
https://mlg.eng.cam.ac.uk/pub/pdf/HusLac11.pdf13 Feb 2023: We foundthat kst outperformed both competing methods sig-nificantly (p < 0.01, two sample T-test). ... 3. As measureof performance, we used average confusion, i. e. thefraction of misclassified test images. -
Recent Advanced in Causal Modelling Using Directed Graphs
https://mlg.eng.cam.ac.uk/zoubin/SALD/scheines.ppt27 Jan 2023: Can be applied to distributions where tests of conditional independence are known, but scores aren’t. ... 2) X1 - X2 into X2. Test. Test Conditions. X3. X2. -
PILCO: A Model-Based and Data-Efficient Approach to Policy Search
https://mlg.eng.cam.ac.uk/pub/pdf/DeiRas11.pdf13 Feb 2023: 3)–(5). Doing this properlyrequires mapping uncertain test inputs through theGP dynamics model. ... b) Histogram (after 1,000 test runs)of the distances of the flywheel frombeing upright. -
The Random Forest Kerneland creating other kernels for big data from…
https://mlg.eng.cam.ac.uk/pub/pdf/DavGha14a.pdf13 Feb 2023: the test log-likelihood on a symmetric logscale as the discrepancy is very large. ... Test MSE on real-world regression problems (lower isbetter). 5.2. Approximation quality. -
Gaussian Mixture Modeling with Gaussian Process Latent Variable Models
https://mlg.eng.cam.ac.uk/pub/pdf/NicRas10.pdf13 Feb 2023: Secondly, our empirical results (see Section 3)indicate, that the test set performance is simply not good. ... Table 1. Average log test densities over 10 random splits of the data. -
Prediction on Spike DataUsing Kernel Algorithms Jan Eichhorn, Andreas …
https://mlg.eng.cam.ac.uk/pub/pdf/EicTolZieetal04.pdf13 Feb 2023: Finallywe train the best model on these four folds and compute an independent test error on theremaining fold. ... Table 1 Mean test error and standard error on the low contrast dataset. -
Nonparametric Bayesian Sparse Factor Models with application to Gene…
https://mlg.eng.cam.ac.uk/pub/pdf/KnoGha11b.pdf13 Feb 2023: Figure 8(a) shows test set log likelihoods for 10 ran-dom divisions of the data into training and test sets. ... test. dat. a. SFAAFA NS. FA. Fig 9. Test set log likelihoods on Prostate cancer dataset from Yu et al. -
Inferring a measure of physiological age frommultiple ageing related…
https://mlg.eng.cam.ac.uk/pub/pdf/KnoParGlaWin11.pdf13 Feb 2023: Table 2: Hold out test. Values are log10(p) where p is the p-value for the Spearman rank correlationhypothesis test. ... The results are shown in Figure 2.2, where we are also able to include binary variables unlikefor the Spearman test. -
Scaling the Indian Buffet Process via Submodular Maximization
https://mlg.eng.cam.ac.uk/pub/pdf/ReeGha13a.pdf13 Feb 2023: aibp. t-aibp. f-vibpi-vibp. seconds. test. log-likelihood ugibbs. t-ugibbs. t-aibp. aibp. bnmf. ... test. log-likelihood. Piano. meibp. ugibbs. aibp. bnmf. f-vibp. i-vibp. 103 104 1054.95. -
MODEL BASED LEARNING OF SIGMA POINTS IN UNSCENTED KALMAN ...
https://mlg.eng.cam.ac.uk/pub/pdf/TurRas10.pdf13 Feb 2023: 11). 2If we want to integrate the parameters out we must run the UKF witheach sample of θ|y1:T during test and average. ... 6.5. Computational Complexity. The UKF-L, UKF, and EKF have test set computationaltime O(DT(D2 M)). -
PROPAGATION OF UNCERTAINTY IN BAYESIAN KERNEL MODELS— APPLICATION TO…
https://mlg.eng.cam.ac.uk/pub/pdf/QuiGirLarRas03.pdf13 Feb 2023: Thiscorresponds to using the model in recall/test phase under uncer-tain input. ... 3. PREDICTION WITH UNCERTAIN INPUT. Assume that the test inputx can not be observed directly and theuncertainty is modeled asx p(x) = N (u, S), with meanuand covariance -
Variable noise and dimensionality reduction forsparse Gaussian…
https://mlg.eng.cam.ac.uk/zoubin/papers/snelson_uai.pdf27 Jan 2023: We have triedto implement both versions efficiently. Validation Time /s. Method NLPD MSE Train Test. ... To test this weused PCA to reduce the dimension to 5, before usingthe SPGP. -
4F13 Machine Learning: Coursework #4: Reinforcement Learning Zoubin…
https://mlg.eng.cam.ac.uk/teaching/4f13/0910/cw/coursework4.pdf19 Nov 2023: Test yourvalueIteration algorithm. -
Gaussian Process Change Point Models
https://mlg.eng.cam.ac.uk/pub/pdf/SaaTurRas10.pdf13 Feb 2023: Weevaluated the models’ ability to predict next day snow-fall using 35 years of test data. ... Method Negative Log Likelihood p-value MSE p-valueNile Data (200 Training Points, 462 Test Points). -
Adaptive Sequential Bayesian Change Point Detection Ryan…
https://mlg.eng.cam.ac.uk/pub/pdf/TurSaaRas09.pdf13 Feb 2023: We also include the 95% error bars on the NLL and the p-value that the joint model/learned hypers hasa higher NLL using a one sided t-test. ... Industry: We test on the last 8455 points of the portfolio data, 3 July 1975–31 December 2008. -
12 October N+Vs layout ok
https://mlg.eng.cam.ac.uk/zoubin/papers/NewsViews00.pdf27 Jan 2023: 1b). results from tests of this material in elec-trolyte cells have shown an acceptable life-time, with a competitive energy per unitweight at reasonable rates of charge and discharge. ... To test this possibility, Keller et al.1 stud-ied three weed -
4F13 Probabilistic Machine Learning: Coursework #1: Gaussian…
https://mlg.eng.cam.ac.uk/teaching/4f13/1617/cw/coursework1.pdf19 Nov 2023: Show and comment on the fit and the hypers, and the predictions forthe test data.
Refine your results
Date
- 314 2023
- 90 Past year
- 55 Past 6 months
- 55 2024
- 50 Past 3 months
- 50 Past month
- 12 Past fortnight
- 6 Past week
- 1 2022
Search history
Recently clicked results
Recently clicked results
Your click history is empty.
Recent searches
Recent searches
Your search history is empty.