Search Results for author: Maciej Skorski

Found 12 papers, 1 papers with code

Mean-Squared Accuracy of Good-Turing Estimator

no code implementations14 Apr 2021 Maciej Skorski

This work characterizes the maximal mean-squared error of the Good-Turing estimator, for any sample \emph{and} alphabet size.

Language Modelling

Confidence-Optimal Random Embeddings

1 code implementation6 Apr 2021 Maciej Skorski

The seminal result of Johnson and Lindenstrauss on random embeddings has been intensively studied in applied and theoretical computer science.

Bernstein-Type Bounds for Beta Distribution

no code implementations6 Jan 2021 Maciej Skorski

This work establishes Bernstein-type closed-form concentration inequalities for the beta distribution, with optimal variance proxy.

Probability Statistics Theory Applications Statistics Theory

Random Embeddings with Optimal Accuracy

no code implementations31 Dec 2020 Maciej Skorski

This work constructs Jonson-Lindenstrauss embeddings with best accuracy, as measured by variance, mean-squared error and exponential concentration of the length distortion.

A Modern Analysis of Hutchinson's Trace Estimator

no code implementations23 Dec 2020 Maciej Skorski

The paper establishes the new state-of-art in the accuracy analysis of Hutchinson's trace estimator.

Simple Analysis of Johnson-Lindenstrauss Transform under Neuroscience Constraints

no code implementations20 Aug 2020 Maciej Skorski

The paper re-analyzes a version of the celebrated Johnson-Lindenstrauss Lemma, in which matrices are subjected to constraints that naturally emerge from neuroscience applications: a) sparsity and b) sign-consistency.

Revisiting Concentration of Missing Mass

no code implementations19 May 2020 Maciej Skorski

We revisit the problem of \emph{missing mass concentration}, developing a new method of estimating concentration of heterogenic sums, in spirit of celebrated Rosenthal's inequality.

Revisiting Initialization of Neural Networks

no code implementations20 Apr 2020 Maciej Skorski, Alessandro Temperoni, Martin Theobald

The proper initialization of weights is crucial for the effective training and fast convergence of deep neural networks (DNNs).

Image Classification

Efficient Sampled Softmax for Tensorflow

no code implementations10 Apr 2020 Maciej Skorski

This short paper discusses an efficient implementation of \emph{sampled softmax loss} for Tensorflow.

Bounds on Bayes Factors for Binomial A/B Testing

no code implementations28 Feb 2019 Maciej Skorski

Bayes factors, in many cases, have been proven to bridge the classic -value based significance testing and bayesian analysis of posterior odds.

Kernel Density Estimation Bias under Minimal Assumptions

no code implementations2 Jan 2019 Maciej Skorski

The contribution of this paper is twofold (a) we demonstrate that, when the bandwidth is an arbitrary invertible matrix going to zero, it is necessary to keep a certain balance between the \emph{kernel decay} and \emph{magnitudes of bandwidth eigenvalues}; in fact, without the sufficient decay the estimates may not be even bounded (b) we give a rigorous derivation of bounds with explicit constants for the bias, under possibly minimal assumptions.

Density Estimation

Simple Root Cause Analysis by Separable Likelihoods

no code implementations13 Aug 2018 Maciej Skorski

Root Cause Analysis for Anomalies is challenging because of the trade-off between the accuracy and its explanatory friendliness, required for industrial applications.

Cannot find the paper you are looking for? You can Submit a new open access paper.