Sparse Dimensionality Reduction Revisited

Mikael Møller Høgsgaard, Lior Kamma, Kasper Green Larsen, Jelani Nelson, Chris Schwiegelshohn
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:18454-18469, 2024.

Abstract

The sparse Johnson-Lindenstrauss transform is one of the central techniques in dimensionality reduction. It supports embedding a set of $n$ points in $\mathbb{R}^d$ into $m=O(\varepsilon^{-2} \ln n)$ dimensions while preserving all pairwise distances to within $1 \pm \varepsilon$. Each input point $x$ is embedded to $Ax$, where $A$ is an $m \times d$ matrix having $s$ non-zeros per column, allowing for an embedding time of $O(s \|x\|_0)$. Since the sparsity of $A$ governs the embedding time, much work has gone into improving the sparsity $s$. The current state-of-the-art by Kane and Nelson (2014) shows that $s = O(\varepsilon^{-1} \ln n)$ suffices. This is almost matched by a lower bound of $s = \Omega(\varepsilon^{-1} \ln n/\ln(1/\varepsilon))$ by Nelson and Nguyen (2013) for $d=\Omega(n)$. Previous work thus suggests that we have near-optimal embeddings. In this work, we revisit sparse embeddings and present a sparser embedding for instances in which $d = n^{o(1)}$, which in many applications is realistic. Formally, our embedding achieves $s = O(\varepsilon^{-1}(\ln n/\ln(1/\varepsilon)+\ln^{2/3}n \ln^{1/3} d))$. We also complement our analysis by strengthening the lower bound of Nelson and Nguyen to hold also when $d \ll n$, thereby matching the first term in our new sparsity upper bound. Finally, we also improve the sparsity of the best oblivious subspace embeddings for optimal embedding dimensionality.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-hogsgaard24a, title = {Sparse Dimensionality Reduction Revisited}, author = {H{\o}gsgaard, Mikael M{\o}ller and Kamma, Lior and Larsen, Kasper Green and Nelson, Jelani and Schwiegelshohn, Chris}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {18454--18469}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/hogsgaard24a/hogsgaard24a.pdf}, url = {https://proceedings.mlr.press/v235/hogsgaard24a.html}, abstract = {The sparse Johnson-Lindenstrauss transform is one of the central techniques in dimensionality reduction. It supports embedding a set of $n$ points in $\mathbb{R}^d$ into $m=O(\varepsilon^{-2} \ln n)$ dimensions while preserving all pairwise distances to within $1 \pm \varepsilon$. Each input point $x$ is embedded to $Ax$, where $A$ is an $m \times d$ matrix having $s$ non-zeros per column, allowing for an embedding time of $O(s \|x\|_0)$. Since the sparsity of $A$ governs the embedding time, much work has gone into improving the sparsity $s$. The current state-of-the-art by Kane and Nelson (2014) shows that $s = O(\varepsilon^{-1} \ln n)$ suffices. This is almost matched by a lower bound of $s = \Omega(\varepsilon^{-1} \ln n/\ln(1/\varepsilon))$ by Nelson and Nguyen (2013) for $d=\Omega(n)$. Previous work thus suggests that we have near-optimal embeddings. In this work, we revisit sparse embeddings and present a sparser embedding for instances in which $d = n^{o(1)}$, which in many applications is realistic. Formally, our embedding achieves $s = O(\varepsilon^{-1}(\ln n/\ln(1/\varepsilon)+\ln^{2/3}n \ln^{1/3} d))$. We also complement our analysis by strengthening the lower bound of Nelson and Nguyen to hold also when $d \ll n$, thereby matching the first term in our new sparsity upper bound. Finally, we also improve the sparsity of the best oblivious subspace embeddings for optimal embedding dimensionality.} }
Endnote
%0 Conference Paper %T Sparse Dimensionality Reduction Revisited %A Mikael Møller Høgsgaard %A Lior Kamma %A Kasper Green Larsen %A Jelani Nelson %A Chris Schwiegelshohn %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-hogsgaard24a %I PMLR %P 18454--18469 %U https://proceedings.mlr.press/v235/hogsgaard24a.html %V 235 %X The sparse Johnson-Lindenstrauss transform is one of the central techniques in dimensionality reduction. It supports embedding a set of $n$ points in $\mathbb{R}^d$ into $m=O(\varepsilon^{-2} \ln n)$ dimensions while preserving all pairwise distances to within $1 \pm \varepsilon$. Each input point $x$ is embedded to $Ax$, where $A$ is an $m \times d$ matrix having $s$ non-zeros per column, allowing for an embedding time of $O(s \|x\|_0)$. Since the sparsity of $A$ governs the embedding time, much work has gone into improving the sparsity $s$. The current state-of-the-art by Kane and Nelson (2014) shows that $s = O(\varepsilon^{-1} \ln n)$ suffices. This is almost matched by a lower bound of $s = \Omega(\varepsilon^{-1} \ln n/\ln(1/\varepsilon))$ by Nelson and Nguyen (2013) for $d=\Omega(n)$. Previous work thus suggests that we have near-optimal embeddings. In this work, we revisit sparse embeddings and present a sparser embedding for instances in which $d = n^{o(1)}$, which in many applications is realistic. Formally, our embedding achieves $s = O(\varepsilon^{-1}(\ln n/\ln(1/\varepsilon)+\ln^{2/3}n \ln^{1/3} d))$. We also complement our analysis by strengthening the lower bound of Nelson and Nguyen to hold also when $d \ll n$, thereby matching the first term in our new sparsity upper bound. Finally, we also improve the sparsity of the best oblivious subspace embeddings for optimal embedding dimensionality.
APA
Høgsgaard, M.M., Kamma, L., Larsen, K.G., Nelson, J. & Schwiegelshohn, C.. (2024). Sparse Dimensionality Reduction Revisited. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:18454-18469 Available from https://proceedings.mlr.press/v235/hogsgaard24a.html.

Related Material