Institution
Company•Tel Aviv, Israel•
About: Facebook is a company organization based out in Tel Aviv, Israel. It is known for research contribution in the topics: Artificial neural network & Language model. The organization has 7856 authors who have published 10906 publications receiving 570123 citations. The organization is also known as: facebook.com & FB.
Topics: Artificial neural network, Language model, Reinforcement learning, Machine translation, Social network
Papers published on a yearly basis
Papers
More filters
••
TL;DR: HuBERT as mentioned in this paper utilizes an offline clustering step to provide aligned target labels for a BERT-like prediction loss, which forces the model to learn a combined acoustic and language model over the continuous inputs.
Abstract: Self-supervised approaches for speech representation learning are challenged by three unique problems: (1) there are multiple sound units in each input utterance, (2) there is no lexicon of input sound units during the pre-training phase, and (3) sound units have variable lengths with no explicit segmentation. To deal with these three problems, we propose the Hidden-Unit BERT (HuBERT) approach for self-supervised speech representation learning, which utilizes an offline clustering step to provide aligned target labels for a BERT-like prediction loss. A key ingredient of our approach is applying the prediction loss over the masked regions only, which forces the model to learn a combined acoustic and language model over the continuous inputs. HuBERT relies primarily on the consistency of the unsupervised clustering step rather than the intrinsic quality of the assigned cluster labels. Starting with a simple k-means teacher of 100 clusters, and using two iterations of clustering, the HuBERT model either matches or improves upon the state-of-the-art wav2vec 2.0 performance on the Librispeech (960 h) and Libri-light (60,000 h) benchmarks with 10 min, 1 h, 10 h, 100 h, and 960 h fine-tuning subsets. Using a 1B parameter model, HuBERT shows up to 19% and 13% relative WER reduction on the more challenging dev-other and test-other evaluation subsets.
1 2
266 citations
•
TL;DR: This article propose an extension to neural network language models to adapt their prediction to the recent history, which stores past hidden activations as memory and accesses them through a dot product with the current hidden activation.
Abstract: We propose an extension to neural network language models to adapt their prediction to the recent history. Our model is a simplified version of memory augmented networks, which stores past hidden activations as memory and accesses them through a dot product with the current hidden activation. This mechanism is very efficient and scales to very large memory sizes. We also draw a link between the use of external memory in neural network and cache models used with count based language models. We demonstrate on several language model datasets that our approach performs significantly better than recent memory augmented networks.
264 citations
••
03 May 2019TL;DR: It is shown that by scaling on various axes (including data size and problem 'hardness'), one can largely match or even exceed the performance of supervised pre-training on a variety of tasks such as object detection, surface normal estimation and visual navigation using reinforcement learning.
Abstract: Self-supervised learning aims to learn representations from the data itself without explicit manual supervision. Existing efforts ignore a crucial aspect of self-supervised learning - the ability to scale to large amount of data because self-supervision requires no manual labels. In this work, we revisit this principle and scale two popular self-supervised approaches to 100 million images. We show that by scaling on various axes (including data size and problem 'hardness'), one can largely match or even exceed the performance of supervised pre-training on a variety of tasks such as object detection, surface normal estimation (3D) and visual navigation using reinforcement learning. Scaling these methods also provides many interesting insights into the limitations of current self-supervised techniques and evaluations. We conclude that current self-supervised methods are not 'hard' enough to take full advantage of large scale data and do not seem to learn effective high level semantic representations. We also introduce an extensive benchmark across 9 different datasets and tasks. We believe that such a benchmark along with comparable evaluation settings is necessary to make meaningful progress. Code is at: https://github.com/facebookresearch/fair_self_supervision_benchmark.
264 citations
••
01 Jan 2017TL;DR: E-UTRAN access technology is also referred to long-term evolution (LTE) with its spectral efficiency increased 150 times as compared to the first-generation analog access technology.
Abstract: Mobile wireless network has evolved from the first generation of analog communications to the fourth-generation evolved universal terrestrial radio access network (E-UTRAN). E-UTRAN access technology is also referred to long-term evolution (LTE) with its spectral efficiency increased 150 times as compared to the first-generation analog access technology.
264 citations
•
19 Mar 2009TL;DR: An analysis of Facebook diffusion chains using zero-inflated negative binomial regressions shows that after controlling for distribution effects, there is no meaningful evidence that a start node’s maximum diffusion chain length can be predicted with the user's demographics or Facebook usage characteristics.
Abstract: Whether they are modeling bookmarking behavior in Flickr or cascades of failure in large networks, models of diffusion often start with the assumption that a few nodes start long chain reactions, resulting in large-scale cascades. While reasonable under some conditions, this assumption may not hold for social media networks, where user engagement is high and information may enter a system from multiple disconnected sources. Using a dataset of 262,985 Facebook Pages and their associated fans, this paper provides an empirical investigation of diffusion through a large social media network. Although Facebook diffusion chains are often extremely long (chains of up to 82 levels have been observed), they are not usually the result of a single chain-reaction event. Rather, these diffusion chains are typically started by a substantial number of users. Large clusters emerge when hundreds or even thousands of short diffusion chains merge together. This paper presents an analysis of these diffusion chains using zero-inflated negative binomial regressions. We show that after controlling for distribution effects, there is no meaningful evidence that a start node’s maximum diffusion chain length can be predicted with the user's demographics or Facebook usage characteristics (including the user's number of Facebook friends). This may provide insight into future research on public opinion formation.
263 citations
Authors
Showing all 7875 results
Name | H-index | Papers | Citations |
---|---|---|---|
Yoshua Bengio | 202 | 1033 | 420313 |
Xiang Zhang | 154 | 1733 | 117576 |
Jitendra Malik | 151 | 493 | 165087 |
Trevor Darrell | 148 | 678 | 181113 |
Christopher D. Manning | 138 | 499 | 147595 |
Robert W. Heath | 128 | 1049 | 73171 |
Pieter Abbeel | 126 | 589 | 70911 |
Yann LeCun | 121 | 369 | 171211 |
Li Fei-Fei | 120 | 420 | 145574 |
Jon Kleinberg | 117 | 444 | 87865 |
Sergey Levine | 115 | 652 | 59769 |
Richard Szeliski | 113 | 359 | 72019 |
Sanjeev Kumar | 113 | 1325 | 54386 |
Bruce Neal | 108 | 561 | 87213 |
Larry S. Davis | 107 | 693 | 49714 |