Eylem Seç
Reliable Online Prediction with Refuse Option
Başlık:
Reliable Online Prediction with Refuse Option
Yazar:
Koçak, Mustafa Anil, author.
ISBN:
9780355981407
Yazar Ek Girişi:
Fiziksel Tanımlama:
1 electronic resource (102 pages)
Genel Not:
Source: Dissertation Abstracts International, Volume: 79-10(E), Section: B.
Advisors: Elza Erkip; Dennis E. Shasha Committee members: Sundeep Rangan.
Özet:
Prediction of the upcoming points in a data stream, online prediction , is a fundamental problem in machine learning. In this thesis, we design meta-algorithms that can take any standard online prediction algorithm and provably improve its performance, given the right to refuse to make predictions in some instances. Allowing refusals means that the meta-algorithm may decline to emit a prediction produced by the base algorithm on occasion. We investigate this problem in two different settings.
First, we assume the data points in the stream are sampled from an exchangeable distribution. Under this assumption, we introduce Conjugate Predictors as a refusing meta-algorithm which guarantees that the error probability of any standard machine learning algorithm is kept below a pre-specified target epsilon. Our approach, based on conformal predictors, refuses less often than state-of-art meta-algorithms, with the level of improvement depending on the characteristics of the base algorithm. We validate our theoretical guarantees and the effectiveness of our approach through experiments on standard machine learning data sets and algorithms, such as k-nearest neighbors and random forests.
Next, we relax assumptions on the distribution of data and propose a novel meta-algorithm, SafePredict, that works with any base predictor for online data. SafePredict guarantees an arbitrarily chosen error rate, epsilon, on non-refused data points. The SafePredict error bound does not rely on any assumptions on the data distribution or the base predictor. When the base predictor happens not to exceed the target error rate epsilon, SafePredict refuses only a finite number of times. When the error rate of the base predictor changes through time, SafePredict makes use of a weight-shifting heuristic that adapts to these changes without knowing when the changes occur yet still maintains the correctness guarantee. Empirical results show that (i) SafePredict compares favorably with state-of-the-art confidence based refusal mechanisms which fail to offer robust error guarantees, and (ii) combining SafePredict with such refusal mechanisms can in many cases further reduce the number of refusals.
Notlar:
School code: 1988
Tüzel Kişi Ek Girişi:
Mevcut:*
Yer Numarası | Demirbaş Numarası | Shelf Location | Lokasyon / Statüsü / İade Tarihi |
---|---|---|---|
XX(678048.1) | 678048-1001 | Proquest E-Tez Koleksiyonu | Arıyor... |
On Order
Liste seç
Bunu varsayılan liste yap.
Öğeler başarıyla eklendi
Öğeler eklenirken hata oldu. Lütfen tekrar deneyiniz.
:
Select An Item
Data usage warning: You will receive one text message for each title you selected.
Standard text messaging rates apply.