English
Related papers

Related papers: On a new notion of regularizer

200 papers

Prior knowledge has been shown very useful to address many natural language processing tasks. Many approaches have been proposed to formalise a variety of knowledge, however, whether the proposed approach is robust or sensitive to the…

Computation and Language · Computer Science 2015-03-04 Biao Liu , Minlie Huang

Estimating the regular normal cone to constraint systems plays an important role for the derivation of sharp necessary optimality conditions. We present two novel approaches and introduce a new stationarity concept which is stronger than…

Optimization and Control · Mathematics 2019-02-21 Matúš Benko , Helmut Gfrerer

Why do large neural network generalize so well on complex tasks such as image classification or speech recognition? What exactly is the role regularization for them? These are arguably among the most important open questions in machine…

Machine Learning · Statistics 2017-04-10 Pirmin Lemberger

Visualization tools for supervised learning have allowed users to interpret, introspect, and gain intuition for the successes and failures of their models. While reinforcement learning practitioners ask many of the same questions, existing…

Machine Learning · Computer Science 2020-07-14 Shuby Deshpande , Jeff Schneider

We introduce the notion of syzygy for a set of reduction operators and relate it to the notion of syzygy for presentations of algebras. We give a method for constructing a linear basis of the space of syzygies for a set of reduction…

Rings and Algebras · Mathematics 2018-04-10 Cyrille Chenavier

Classification may not be reliable for several reasons: noise in the data, insufficient input information, overlapping distributions and sharp definition of classes. Faced with several possibilities neural network may in such cases still be…

Machine Learning · Computer Science 2019-01-29 Włodzisław Duch , Rafał Adamczak , Yoichi Hayashi

In this paper, we determine the complexity of the satisfiability problem for various logics obtained by adding numerical quantifiers, and other constructions, to the traditional syllogistic. In addition, we demonstrate the incompleteness of…

Logic in Computer Science · Computer Science 2024-04-19 Ian Pratt-Hartmann

This essay considers the special character of mathematical reasoning, and draws on observations from interactive theorem proving and the history of mathematics to clarify the nature of formal and informal mathematical language. It proposes…

History and Overview · Mathematics 2015-08-24 Jeremy Avigad

We introduce the structural resource lambda-calculus, a new formalism in which strongly normalizing terms of the lambda-calculus can naturally be represented, and at the same time any type derivation can be internally rewritten to its…

Logic in Computer Science · Computer Science 2025-03-26 Ugo Dal Lago , Federico Olimpieri

Overfitting is one of the most critical challenges in deep neural networks, and there are various types of regularization methods to improve generalization performance. Injecting noises to hidden units during training, e.g., dropout, is…

Machine Learning · Computer Science 2017-11-10 Hyeonwoo Noh , Tackgeun You , Jonghwan Mun , Bohyung Han

Most of the work on interpretable machine learning has focused on designing either inherently interpretable models, which typically trade-off accuracy for interpretability, or post-hoc explanation systems, which lack guarantees about their…

Machine Learning · Computer Science 2019-06-05 Gregory Plumb , Maruan Al-Shedivat , Eric Xing , Ameet Talwalkar

The use of convex regularizers allows for easy optimization, though they often produce biased estimation and inferior prediction performance. Recently, nonconvex regularizers have attracted a lot of attention and outperformed convex ones.…

Optimization and Control · Mathematics 2017-02-14 Quanming Yao , James. T Kwok

Convenient parameterizations of matrices in terms of vectors transform (certain classes of) matrix equations into covariant (hence rotation-invariant) vector equations. Certain recently introduced such parameterizations are tersely…

Exactly Solvable and Integrable Systems · Physics 2009-11-10 M. Bruschi , F. Calogero

We introduce the notion of an approximation system as a generalization of Taylor approximation, and we give some first examples. Next we develop the general theory, including error bounds and a sufficient criterion for convergence. More…

Classical Analysis and ODEs · Mathematics 2017-10-20 Victor A. Pessers , Tom H. Koornwinder

The aim of this paper is to numerically study the performance of a method of regularization. This technique was developed to solve the illposed problem of estimating a source-dimensional Poisson equation for two dimensions from measurements…

Analysis of PDEs · Mathematics 2024-04-22 Guillermo Federico Umbricht , Diana Rubio

The proposed system of integer functions is logically fully independent from the traditional mathematical analysis of the real functions, but there is a well-defined mutual correspondence between the two disciplines. The system of integer…

General Mathematics · Mathematics 2017-10-03 Jozsef Peredy

Main purposes of the paper are followings: 1) To show examples of the calculations in domain of QFT via ``derivative rules'' of an expert system; 2) To consider advantages and disadvantage that technology of the calculations; 3) To reflect…

Artificial Intelligence · Computer Science 2007-05-23 Andrea Severe

Logically constrained term rewriting is a relatively new formalism where rules are equipped with constraints over some arbitrary theory. Although there are many recent advances with respect to rewriting induction, completion, complexity…

Logic in Computer Science · Computer Science 2024-07-08 Takahito Aoto , Naoki Nishida , Jonas Schöpf

This paper introduces to the calculus of regular expressions a replace operator, ->, and defines a set of replacement expressions that concisely encode several alternate variations of the operation. The basic case is unconditional…

cmp-lg · Computer Science 2008-02-03 Lauri Karttunen , Rank Xerox Research Centre

Regularization and interior point approaches offer valuable perspectives to address constrained nonlinear optimization problems in view of control applications. This paper discusses the interactions between these techniques and proposes an…

Optimization and Control · Mathematics 2022-10-31 Alberto De Marchi