Facebook AI Is Teaching Robots to Perceive, Understand, and Interact Through Touch
On November 1, Facebook AI Research shared its progress on developing AI systems that can understand and interact through touch.
Tag: Facebook AI Research
Yann LeCun Team Uses Dictionary Learning To Peek Into Transformers’ Black Boxes
A Yann LeCun team proposes dictionary learning to provide detailed visualizations of transformer representations and insights into semantic structures such as word-level disambiguation, sentence-level pattern formation, and long-range dependency captured by transformers.
Meet CSL: Facebook AI Approach Achieves 19% WER Reduction for ASR Under Ultra Low-Resource Conditions
A research team from Facebook AI proposes a Contrastive Semi-supervised Learning (CSL) approach that synthesizes pseudo-labelling and contrastive losses to improve the stability of learned speech representations for automatic speech recognition (ASR).
Yann LeCun Hails MSA Transformer’s ‘Huge Progress’ in Protein Contact Prediction
UC Berkeley, Facebook AI Research and New York University researchers’ Multiple Sequence Alignments (MSA) Transformer surpasses current state-of-the-art unsupervised structure learning methods by a wide margin.
VisualVoice Uses Facial Appearance to Boost SOTA in Speech Separation
Recent AI research on speech separation has explored ways to associate lip motions in videos with audio, but this approach suffers when speakers’ lips are occluded, which they often are in busy multi-speaker environments.
Facebook AI Says Class Selectivity In Neurons May Impair DNN Performance
Facebook AI says DNNs can perform well without class specific neurons and overreliance on intuition-based methods for understanding DNNs can be misleading.
Facebook & CMU Introduce TaBERT for Understanding Tabular Data Queries
TaBERT-powered neural semantic parsers showed performance improvements on the challenging benchmark WikiTableQuestions and demonstrated competitive performance on the text-to-SQL dataset Spider.
Researchers Propose ‘Neuro-Symbolic’ Approach for Generative Art
Researchers propose a neuro-symbolic hybrid approach to address the challenge of creativity in generative art.
Facebook AI’s DETR Applies Transformers to CV Tasks
Facebook this week released Detection Transformers (DETR), a new approach for object detection and panoptic segmentation tasks that uses a completely different architecture than previous object detection systems.
Facebook’s Highly Efficient New Real-Time Text-To-Speech System Runs on CPUs
To deliver human-level voices to its platform’s billions of users while maintaining strict compute efficiency, Facebook AI researchers have deployed a new neural TTS system that works on CPU servers.
‘Active Neural Slam’ Uses Classical and Learning Approaches to Explore 3D Spaces
Researchers have introduced Active Neural SLAM, a modular and hierarchical approach to learning policies for exploring 3D environments.
Use the Force! AI Predicts Human-Object Contact Points and Forces From Video
In a new study, researchers use a physics simulator to learn to predict physical forces in videos of humans interacting with objects.
Radioactive Data: Facebook AI Knows Where You Got Your Training Dataset
Researchers proposed a “radioactive data” technique for subtly marking images in a dataset to help researchers later determine whether they were used to train a particular model.
Facebook HiPlot ‘Makes Understanding High Dimensional Data Easy’
Facebook’s new HiPlot is a lightweight interactive visualization tool that takes this further, using parallel plots to discover correlations and patterns in such high-dimensional data.
Facebook AI mBART: The Tower of Babel’s Silicon Solution
Facebook AI researchers have further developed the BART model with the introduction of mBART.
Facebook PointRend: Rendering Image Segmentation
Facebook AI Research team has introduced a new “point-based rendering” neural network module with an iterative subdivision algorithm that can integrate SOTA image segmentation models.
12-in-1: Facebook AI’s New Framework Tackles Multiple Vision-and-Language Tasks
The model reduces the number of parameters from some 3 billion to 270 million while improving task performance by an average of 2.05 points.
2020 AI Residency Guide
As global AI development and deployment continues, the demand for AI talents is growing faster than ever. A number of industry leaders and reputable institutions offer AI residency programs designed to help nurture promising AI talents.
Can Pretrained Language Models Replace Knowledge Bases?
The recent rapid development of pretrained language models has produced significant performance improvements on downstream NLP tasks.
Facebook AI Memory Layer Boosts Network Capacity by a Billion Parameters
Recently, Facebook AI Research (FAIR) researchers introduced a structured memory layer which can be easily integrated into a neural network to greatly expand network capacity and the number of parameters without significantly changing calculation cost.
From Foodie Pic to Your Plate: Generating Recipes With Facebook AI
acebook’s AI’s new “Inverse Cooking” AI system reverse-engineers recipes from food images, predicting both the ingredients in the dish and their preparation and cooking instructions.
MNIST Reborn, Restored and Expanded: Additional 50K Training Samples
Researchers from New York University and Facebook AI Research recently added 50,000 test samples to the dataset. Facebook Chief AI Scientist Yann LeCun, who co-developed the MNIST, tweeted his approval: “MNIST reborn, restored and expanded.”
‘Hot’ Papers from Google Brain, DeepMind and Facebook AI
Synced Global AI Weekly June 2nd
FAIR Paper Questions Pre-Training’s Efficacy in CV
ImageNet Pre-training is common in a variety of CV (Computer Vision) tasks, as something of a consensus has emerged that pre-training can help a model learn transferrable information that can be useful for target tasks.