What is machine learning / ai? How to lean machine learning in practice? There are a lot of interested people out there but many do not know where to start. The difficult question basically is how to start actually learning it? Especially beginners might get discouraged because of statistics and math which is an integral part of machine learning.

Researchers from North Carolina State University have discovered that teaching physics to neural networks enables those networks to better adapt to chaos within their environment. The work has implications for improved artificial intelligence (AI) applications ranging from medical diagnostics to automated drone piloting. Neural networks are an advanced type of AI loosely based on the way that our brains work. Our natural neurons exchange electrical impulses according to the strengths of their connections. Artificial neural networks mimic this behavior by adjusting numerical weights and biases during training sessions to minimize the difference between their actual and desired outputs.

Earlier this spring, Amazon notified grant applicants that they were recipients of the 2019 Amazon Research Awards, a grant program that provides up to $80,000 in cash and $20,000 in AWS Promotional Credits to academic researchers investigating topics across 11 focus areas. Today, we're publicly announcing the 51 award recipients who represent 39 universities in 10 countries. The 2019 awards averaged $72,000 in cash awards and $15,000 in AWS Promotional Credits in support of each research project. Each grant is intended to support the work of one to two graduate students or postdoctoral students for one year, under the supervision of a faculty member. The 11 focus areas of this year's research awards are computer vision; fairness in artificial intelligence; knowledge management and data quality; machine learning algorithms and theory; natural-language processing; online advertising; operations research and optimization; personalization; robotics; search and information retrieval; and security, privacy, and abuse prevention.

Scientists have created a glove that translates sign language into speech in a matter of seconds through a smartphone app. The hi-tech glove sends electronic signals generated from different movements of the fingers to a circuit board, which are then transmitted wirelessly to the phone. The app translates letters from movements on one hand, as per the American Sign Language (ASL) standard, into phrases, which are uttered by an automated voice. The patented technology aims to reduce the language barrier between people with hearing or speech impediments and those who don't understand sign language. Components used to make the device are cheap, flexible and highly durable, according to the team of engineers in the US, who plan to commercialise the technology if they can speed up the translation time even further.

Number two is Machine Learning by Stanford.

With the development of technology, comes a passing of the torch for programming languages. With historic ties back to Ada Lovelace in 1842, programming has become an essential tool for development in tech, from hand-written assembly language programmes in the 40s to the incredibly developed available today. Gone are the days of Fortran, COBOL, Algol and APL, replaced with many of those listed below. As somewhat of a directory, we have collated the most'in demand' languages in 2020 as well as their difficulty and links to free resources to start your learning journey. Python is an interpreted, object-oriented, high-level programming language with dynamic semantics.

Learning-to-learn (using optimization algorithms to learn a new optimizer) has successfully trained efficient optimizers in practice. This approach relies on meta-gradient descent on a meta-objective based on the trajectory that the optimizer generates. However, there were few theoretical guarantees on how to avoid meta-gradient explosion/vanishing problems, or how to train an optimizer with good generalization performance. In this paper, we study the learning-to-learn approach on a simple problem of tuning the step size for quadratic loss. Our results show that although there is a way to design the meta-objective so that the meta-gradient remain polynomially bounded, computing the meta-gradient directly using backpropagation leads to numerical issues that look similar to gradient explosion/vanishing problems. We also characterize when it is necessary to compute the meta-objective on a separate validation set instead of the original training set. Finally, we verify our results empirically and show that a similar phenomenon appears even for more complicated learned optimizers parametrized by neural networks.

Continuously learning to solve unseen tasks with limited experience has been extensively pursued in meta-learning and continual learning, but with restricted assumptions such as accessible task distributions, independently and identically distributed tasks, and clear task delineations. However, real-world physical tasks frequently violate these assumptions, resulting in performance degradation. This paper proposes a continual online model-based reinforcement learning approach that does not require pre-training to solve task-agnostic problems with unknown task boundaries. We maintain a mixture of experts to handle nonstationarity, and represent each different type of dynamics with a Gaussian Process to efficiently leverage collected data and expressively model uncertainty. We propose a transition prior to account for the temporal dependencies in streaming data and update the mixture online via sequential variational inference. Our approach reliably handles the task distribution shift by generating new models for never-before-seen dynamics and reusing old models for previously seen dynamics. In experiments, our approach outperforms alternative methods in non-stationary tasks, including classic control with changing dynamics and decision making in different driving scenarios.

Robustness issues arise in a variety of forms and are studied through multiple lenses in the machine learning literature. Neural networks lack adversarial robustness -- they are vulnerable to adversarial examples that through small perturbations to inputs cause incorrect predictions. Further, trust is undermined when models give miscalibrated or unstable uncertainty estimates, i.e. the predicted probability is not a good indicator of how much we should trust our model and could vary greatly over multiple independent runs. In this paper, we study the connection between adversarial robustness, predictive uncertainty (calibration) and model uncertainty (stability) on multiple classification networks and datasets. We find that the inputs for which the model is sensitive to small perturbations (are easily attacked) are more likely to have poorly calibrated and unstable predictions. Based on this insight, we examine if calibration and stability can be improved by addressing those adversarially unrobust inputs. To this end, we propose Adversarial Robustness based Adaptive Label Smoothing (AR-AdaLS) that integrates the correlations of adversarial robustness and uncertainty into training by adaptively softening labels conditioned on how easily it can be attacked by adversarial examples. We find that our method, taking the adversarial robustness of the in-distribution data into consideration, leads to better calibration and stability over the model even under distributional shifts. In addition, AR-AdaLS can also be applied to an ensemble model to achieve the best calibration performance.

Manifold Learning occupies a vital role in the field of nonlinear dimensionality reduction and its ideas also serve for other relevant methods. Graph-based methods such as Graph Convolutional Networks (GCN) show ideas in common with manifold learning, although they belong to different fields. Inspired by GCN, we introduce neighbor propagation into LLE and propose Local Neighbor Propagation Embedding (LNPE). With linear computational complexity increase compared with LLE, LNPE enhances the local connections and interactions between neighborhoods by extending $1$-hop neighbors into $n$-hop neighbors. The experimental results show that LNPE could obtain more faithful and robust embeddings with better topological and geometrical properties.