Search Results for author:
Found 13 papers, 4 papers with code
Robust multimodal models have outlier features and encode more concepts
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).
Spatial LibriSpeech: An Augmented Dataset for Spatial Audio Learning
We present Spatial LibriSpeech, a spatial audio dataset with over 650 hours of 19-channel audio, first-order ambisonics, and optional distractor noise.
The Role of Entropy and Reconstruction in Multi-View Self-Supervised Learning
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.
DUET: 2D Structured and Approximately Equivariant Representations
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.
Designing Data: Proactive Data Collection and Iteration for Machine Learning
Lack of diversity in data collection has caused significant failures in machine learning (ML) applications.
Homomorphic Self-Supervised Learning
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.
Contrastive Self-Supervised Learning for Skeleton Representations
Human skeleton point clouds are commonly used to automatically classify and predict the behaviour of others.
Fair SA: Sensitivity Analysis for Fairness in Face Recognition
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.
Challenges of Adversarial Image Augmentations
Image augmentations applied during training are crucial for the generalization performance of image classifiers.
Self-conditioning pre-trained language models
We compare our method with FUDGE and PPLM-BoW, and show that our approach is able to achieve gender parity at a lower perplexity.
Finding Experts in Transformer Models
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.
NETWORK COMPRESSION USING CORRELATION ANALYSIS OF LAYER RESPONSES
Principal Filter Analysis (PFA) is an easy to implement, yet effective method for neural network compression.
Filter Distillation for Network Compression
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.
