no code implementations • 13 Mar 2024 • Tuukka Korhonen, Fedor V. Fomin, Pekka Parviainen
When it comes to the number of tests, our upper bound on the sizes of conditioning sets implies that every $n$-vertex graph can be learned by at most $n^{\kappa}$ tests with conditioning sets of sizes at most $\kappa$.
1 code implementation • 26 Jan 2023 • Hyeongji Kim, Pekka Parviainen, Terje Berge, Ketil Malde
In addition to the standard classification accuracy, we evaluate the hierarchical inference performance by inspecting learned class representatives and the hierarchy-informed performance, i. e., the classification performance, and the metric learning performance by considering predefined hierarchical structures.
no code implementations • 15 Jul 2022 • Johannes Fiedler, Adrià Salvador Palau, Eivind Kristen Osestad, Pekka Parviainen, Bodil Holst
This approximation is then used to generate the initial population of the genetic optimisation algorithm that can converge to arbitrary precision.
1 code implementation • 21 Jan 2022 • Hyeongji Kim, Pekka Parviainen, Ketil Malde
We propose a distance-ratio-based (DR) formulation for metric learning.
no code implementations • 1 Jul 2021 • Pierre Gillot, Pekka Parviainen
Bayesian networks represent relations between variables using a directed acyclic graph (DAG).
1 code implementation • 6 May 2020 • Hyeongji Kim, Pekka Parviainen, Ketil Malde
As a result, adversarial accuracy based on this adversary avoids a tradeoff between accuracy and adversarial accuracy on training data even when $\epsilon$ is large.
no code implementations • 2 Mar 2017 • Xiangju Qin, Paul Blomstedt, Eemeli Leppäaho, Pekka Parviainen, Samuel Kaski
Bayesian matrix factorization (BMF) is a powerful tool for producing low-rank representations of matrices and for predicting missing values and providing confidence intervals.
1 code implementation • NeurIPS 2015 • Janne H. Korhonen, Pekka Parviainen
Both learning and inference tasks on Bayesian networks are NP-hard in general.
no code implementations • 31 Aug 2015 • Pekka Parviainen, Samuel Kaski
We show that for dependency structures between groups to be expressible exactly, the data have to satisfy the so-called groupwise faithfulness assumption.