no code implementations • 19 Oct 2023 • Jonathan Crabbé, Pau Rodríguez, Vaishaal Shankar, Luca Zappella, Arno Blaas
In this work, we bridge this gap by probing the representation spaces of 12 robust multimodal models with various backbones (ResNets and ViTs) and pretraining sets (OpenAI, LAION-400M, LAION-2B, YFCC15M, CC12M and DataComp).
1 code implementation • 18 Aug 2023 • Miguel Sarabia, Elena Menyaylenko, Alessandro Toso, Skyler Seto, Zakaria Aldeneh, Shadi Pirhosseinloo, Luca Zappella, Barry-John Theobald, Nicholas Apostoloff, Jonathan Sheaffer
We present Spatial LibriSpeech, a spatial audio dataset with over 650 hours of 19-channel audio, first-order ambisonics, and optional distractor noise.
1 code implementation • 20 Jul 2023 • Borja Rodríguez-Gálvez, Arno Blaas, Pau Rodríguez, Adam Goliński, Xavier Suau, Jason Ramapuram, Dan Busbridge, Luca Zappella
We consider a different lower bound on the MI consisting of an entropy and a reconstruction term (ER), and analyze the main MVSSL families through its lens.
1 code implementation • 28 Jun 2023 • Xavier Suau, Federico Danieli, T. Anderson Keller, Arno Blaas, Chen Huang, Jason Ramapuram, Dan Busbridge, Luca Zappella
We propose 2D strUctured and EquivarianT representations (coined DUET), which are 2d representations organized in a matrix structure, and equivariant with respect to transformations acting on the input data.
no code implementations • 24 Jan 2023 • Aspen Hopkins, Fred Hohman, Luca Zappella, Xavier Suau Cuadros, Dominik Moritz
Lack of diversity in data collection has caused significant failures in machine learning (ML) applications.
no code implementations • 15 Nov 2022 • T. Anderson Keller, Xavier Suau, Luca Zappella
In this work, we observe that many existing self-supervised learning algorithms can be both unified and generalized when seen through the lens of equivariant representations.
no code implementations • 10 Nov 2022 • Nico Lingg, Miguel Sarabia, Luca Zappella, Barry-John Theobald
Human skeleton point clouds are commonly used to automatically classify and predict the behaviour of others.
no code implementations • 8 Feb 2022 • Aparna R. Joshi, Xavier Suau, Nivedha Sivakumar, Luca Zappella, Nicholas Apostoloff
One such high impact domain is that of face recognition, with real world applications involving images affected by various degradations, such as motion blur or high exposure.
no code implementations • NeurIPS Workshop ICBINB 2021 • Arno Blaas, Xavier Suau, Jason Ramapuram, Nicholas Apostoloff, Luca Zappella
Image augmentations applied during training are crucial for the generalization performance of image classifiers.
1 code implementation • 30 Sep 2021 • Xavier Suau, Luca Zappella, Nicholas Apostoloff
We compare our method with FUDGE and PPLM-BoW, and show that our approach is able to achieve gender parity at a lower perplexity.
no code implementations • 15 May 2020 • Xavier Suau, Luca Zappella, Nicholas Apostoloff
We show that expert units are important in several ways: (1) The presence of expert units is correlated ($r^2=0. 833$) with the generalization power of TM, which allows ranking TM without requiring fine-tuning on suites of downstream tasks.
no code implementations • ICLR 2019 • Xavier Suau, Luca Zappella, Nicholas Apostoloff
Principal Filter Analysis (PFA) is an easy to implement, yet effective method for neural network compression.
no code implementations • ICLR 2019 • Xavier Suau, Luca Zappella, Nicholas Apostoloff
We propose two algorithms: the first allows users to target compression to specific network property, such as number of trainable variable (footprint), and produces a compressed model that satisfies the requested property while preserving the maximum amount of spectral energy in the responses of each layer, while the second is a parameter-free heuristic that selects the compression used at each layer by trying to mimic an ideal set of uncorrelated responses.