Deep learning is seeing tremendous adoption in different industries. One specific area where deep learning has shown great potential is Computer Vision. I personally graduated from a computer vision master's program and went immediately to work in the industry. So what follows is my take on different trends that I am seeing in companies that are using deep learning to tackle challenging computer vision problems. So going back to my studies, in the middle of the master's program, I did an internship in a company in Luxembourg that makes large scanners of wood!

"I didn't think we would get to this point in my lifetime." That's how one research leader in structural biology responded to last week's publication of research in which artificial intelligence (AI) was used to predict the structure of more than 20,000 human proteins, as well as that of nearly all the known proteins produced by 20 model organisms such as Escherichia coli, fruit flies and yeast, but also soya bean and Asian rice. That is a combined total of around 365,000 predictions1. The data, publicly accessible for the first time (see https://alphafold.ebi.ac.uk), were released online on 22 July by researchers at DeepMind, a London-based AI company owned by Google's parent company, Alphabet, and the European Bioinformatics Institute, based at the European Molecular Biology Laboratory (EBI-EMBL) near Cambridge, UK. DeepMind's AI predicts structures for a vast trove of proteins The DeepMind team developed a machine-learning tool called AlphaFold.

Artificial intelligence for IT operations (AIOps) combines sophisticated methods from deep learning, data streaming processing, and domain knowledge to analyse infrastructure data from internal and external sources to automate operations and detect anomalies (unusual system behavior) before they impact the quality of service. Odej Kao, professor at the University of Technology Berlin, gave a keynote presentation about artificial intelligence for IT operations at DevOpsCon Berlin 2021. In data stream processing we frequently struggle to find sufficient amounts of data. On the other hand, in AIOps we have many different sources (e.g., metric, logs, tracing, events, alerts) with several Terabytes of data produced in a typical IT infrastructure per day. We utilize the power of these hidden gems to assist DevOps administrators and jointly with the AI-models improve the availability, security, and the performance of the overall system.

A great deal has been written about the technologies fueling 5G, especially how those technologies will improve the experience that users have regarding connectivity. Similarly, much has been said about how ongoing developments in technology will usher in a new generation of network-aware applications. In this article, we discuss one key aspect of 5G technology and how it will impact the development of wireless network capacity. This is one of the more important but often neglected aspects of wireless communication evolution. It represents yet another important reason why the convergence of cloud computing and wireless communications makes so much sense.

Deep learning has yielded some fantastic results for basic natural language processing (NLP) functions such as named entity recognition (NER), document classification and sentiment analysis -- not to mention the abilities to generate everything from believable short stories to HTML code with minimal text inputs or prompts. In addition, deep learning can also have a dramatic impact on F1 scores, which are used as a performance measure for precision and recall, and so vendors have started to throw much of their weight and resources behind what they see as a game-changing technology. But as the CEO of a company that's been doing NLP for well over 15 years, I don't believe that deep learning is always the answer -- especially from an economic standpoint. I've watched as many new players have stepped up to the plate with NLP solutions underpinned by deep learning. But what I'm not seeing is evidence of big commercial wins, and I suspect that the cost of using deep learning-backed NLP is wiping out significant dollar gains. Deep learning tools like BERT can deliver results, but sometimes at a much greater cost than taking a traditional machine learning approach, depending on the size of your project.

All the sessions from Transform 2021 are available on-demand now. Even before they speak their first words, human babies develop mental models about objects and people. This is one of the key capabilities that allows us humans to learn to live socially and cooperate (or compete) with each other. But for artificial intelligence, even the most basic behavioral reasoning tasks remain a challenge. Advanced deep learning models can do complicated tasks such as detect people and objects in images, sometimes even better than humans.

Climate change has become one of the biggest global problems increasingly compromising the Earth's habitability. Recent developments such as the extraordinary heat waves in California & Canada, and the devastating floods in Germany point to the role of climate change in the ever-increasing frequency of extreme weather. Numerical modelling of the weather and climate have seen tremendous improvements in the last five decades, yet stringent limitations remain to be overcome. Spatially and temporally localized forecasting is the need of the hour for effective adaptation measures towards minimizing the loss of life and property. Artificial Intelligence-based methods are demonstrating promising results in improving predictions, but are still limited by the availability of requisite hardware and software required to process the vast deluge of data at a scale of the planet Earth. Quantum computing is an emerging paradigm that has found potential applicability in several fields. In this opinion piece, we argue that new developments in Artificial Intelligence algorithms designed for quantum computers - also known as Quantum Artificial Intelligence (QAI) - may provide the key breakthroughs necessary to furthering the science of climate change. The resultant improvements in weather and climate forecasts are expected to cascade to numerous societal benefits.

Training and inference in deep neural networks (DNNs) has, due to a steady increase in architectural complexity and data set size, lead to the development of strategies for reducing time and space requirements of DNN training and inference, which is of particular importance in scenarios where training takes place in resource constrained computation environments or inference is part of a time critical application. In this survey, we aim to provide a general overview and categorization of state-of-the-art (SOTA) of techniques to reduced DNN training and inference time and space complexities with a particular focus on architectural adaptions.

After an autoencoder (AE) has learnt to reconstruct one dataset, it might be expected that the likelihood on an out-of-distribution (OOD) input would be low. This has been studied as an approach to detect OOD inputs. Recent work showed this intuitive approach can fail for the dataset pairs FashionMNIST vs MNIST. This paper suggests this is due to the use of Bernoulli likelihood and analyses why this is the case, proposing two fixes: 1) Compute the uncertainty of likelihood estimate by using a Bayesian version of the AE. 2) Use alternative distributions to model the likelihood.

A common lens to theoretically study neural net architectures is to analyze the functions they can approximate. However, the constructions from approximation theory often have unrealistic aspects, for example, reliance on infinite precision to memorize target function values, which make these results potentially less meaningful. To address these issues, this work proposes a formal definition of statistically meaningful approximation which requires the approximating network to exhibit good statistical learnability. We present case studies on statistically meaningful approximation for two classes of functions: boolean circuits and Turing machines. We show that overparameterized feedforward neural nets can statistically meaningfully approximate boolean circuits with sample complexity depending only polynomially on the circuit size, not the size of the approximating network. In addition, we show that transformers can statistically meaningfully approximate Turing machines with computation time bounded by $T$, requiring sample complexity polynomial in the alphabet size, state space size, and $\log (T)$. Our analysis introduces new tools for generalization bounds that provide much tighter sample complexity guarantees than the typical VC-dimension or norm-based bounds, which may be of independent interest.