Machine learning, task automation and robotics are already widely used in business. These and other AI technologies are about to multiply, and we look at how organizations can best take advantage of them. GTC, Nvidia's flagship event, is always a source of announcements around all things AI. The fall 2021 edition is no exception. Omniverse is Nvidia's virtual world simulation and collaboration platform for 3D workflows, bringing its technologies together.

Today, we'll discuss another popular method used to improve the performance of your deep neural network called batch normalization. It is a technique for training deep neural networks that standardizes the inputs to a layer for each mini-batch. After finishing the theoretical part, we will explain how to implement batch normalization in Python using PyTorch. So, let's begin with our lecture. In order to understand batch normalization, first, we need to understand what data normalization is. Data normalization is the process of rescaling the input values in the training dataset to the interval of 0 to 1.

In many applications of machine learning, such as machine learning for medical diagnosis, we would like to have machine learning algorithms that do not memorize sensitive information about the training set, such as the specific medical histories of individual patients. Differential privacy is a notion that allows quantifying the degree of privacy protection provided by an algorithm on the underlying (sensitive) data set it operates on. Through the lens of differential privacy, we can design machine learning algorithms that responsibly train models on private data. Machine learning algorithms work by studying a lot of data and updating their parameters to encode the relationships in that data. Ideally, we would like the parameters of these machine learning models to encode general patterns (e.g., ''patients who smoke are more likely to have heart disease'') rather than facts about specific training examples (e.g., "Jane Smith has heart disease"). Unfortunately, machine learning algorithms do not learn to ignore these specifics by default. If we want to use machine learning to solve an important task, like making a cancer diagnosis model, then when we publish that machine learning model (for example, by making an open source cancer diagnosis model for doctors all over the world to use) we might also inadvertently reveal information about the training set.

Advances in the AI behind speech recognition are driving growth in the market, attracting venture capital and funding startups, posing challenges to established players. The growing acceptance and use of speech recognition devices are driving the market, which according to an estimate by Meticulous Research is expected to reach $26.8 billion globally by 2025, according to a recent account in Analytics Insight. Better speed and accuracy are among the benefits of the evolving technology. One company in the throes of this new growth, AssemblyAI of San Francisco, is offering an API for speech recognition capable of transcribing videos, podcasts, phone calls, and remote meetings. The company was founded by CEO Dylan Fox in 2017 and has received backing from Y Combinator, a startup accelerator, as well as NVIDIA.

Google, Facebook & Amazon all use deep learning methods, but how does it work? In this video Computerphile explains how deep learning works. Research Fellow & Deep Learning Expert Brais Martinez explains. Deep learning (also known as deep structured learning) is part of a broader family of machine learning methods based on artificial neural networks with representation learning. Learning can be supervised, semi-supervised or unsupervised.

Using @tf.function (a way to speed up your regular Python functions)

The new company aims to duplicate the success of DeepMind, another Alphabet subsidiary that used AI to predict the 3D structure of a protein directly from its amino acid sequence. While the announcement was made official on Thursday, it was consolidated in February, according to a filing with Companies House, a U.K. company registry.

Amid the boom of AI in application building, companies face a significant data-labeling problem, especially when it comes to labeling images or other media content they want to train deep learning algorithms on. Today data-labeling and infrastructure provider Scale AI launched a service called Scale Rapid that aims to solve this problem by labeling a data sample within one to three hours. Users can review the work to make sure the labeling is being done correctly, iterate upon their labeling instructions if necessary, and then ramp up to have Scale AI label the rest of their dataset. This is the latest in a series of products Scale AI has launched in the last year as it seeks to maintain its leadership in the labeling sphere. In April, the company raised $325 million, bringing its total raised to over $602 million.

The recent release of GPT-3 got me interested in the state of zero-shot learning and few-shot learning in NLP. While most of the zero-shot learning research is centered around Computer Vision, there has been some interesting work in the NLP domain as well. Over the last few decades, machines have become much more intelligent, but without a properly labeled training data set of seen classes, they cannot distinguish between two similar objects. On the other hand, humans are capable of identifying approximately 30,000 basic object categories. In machine learning, this is considered as the problem of Zero-shot learning (ZSL).

Specifying tasks with videos is a powerful technique towards acquiring novel and general robot skills. However, reasoning over mechanics and dexterous interactions can make it challenging to scale learning contact-rich manipulation. In this work, we focus on the problem of visual non-prehensile planar manipulation: given a video of an object in planar motion, find contact-aware robot actions that reproduce the same object motion. We propose a novel architecture, Differentiable Learning for Manipulation (\ours), that combines video decoding neural models with priors from contact mechanics by leveraging differentiable optimization and finite difference based simulation. Through extensive simulated experiments, we investigate the interplay between traditional model-based techniques and modern deep learning approaches. We find that our modular and fully differentiable architecture performs better than learning-only methods on unseen objects and motions. \url{https://github.com/baceituno/dlm}.