DeepAI AI Chat
Log In Sign Up

List and Certificate Complexities in Replicable Learning

by   Peter Dixon, et al.

We investigate replicable learning algorithms. Ideally, we would like to design algorithms that output the same canonical model over multiple runs, even when different runs observe a different set of samples from the unknown data distribution. In general, such a strong notion of replicability is not achievable. Thus we consider two feasible notions of replicability called list replicability and certificate replicability. Intuitively, these notions capture the degree of (non) replicability. We design algorithms for certain learning problems that are optimal in list and certificate complexity. We establish matching impossibility results.


page 1

page 2

page 3

page 4


Optimal (degree+1)-Coloring in Congested Clique

We consider the distributed complexity of the (degree+1)-list coloring p...

Envy-freeness and Relaxed Stability for Lower-Quotas : A Parameterized Perspective

We consider the problem of assigning agents to resources under the two-s...

A Characterization of List Learnability

A classical result in learning theory shows the equivalence of PAC learn...

Filling the Complexity Gaps for Colouring Planar and Bounded Degree Graphs

We consider a natural restriction of the List Colouring problem: k-Regul...

Joint Design of Convolutional Code and CRC under Serial List Viterbi Decoding

This paper studies the joint design of optimal convolutional codes (CCs)...

Galloping in natural merge sorts

We study the algorithm TimSort and the sub-routine it uses to merge mono...

Weighted and Branching Bisimilarities from Generalized Open Maps

In the open map approach to bisimilarity, the paths and their runs in a ...