Welcome to our October 2021 monthly digest where you can catch up with any AIhub stories you may have missed, get the low-down on recent events, and much more. In this edition we cover our latest focus issue, the concept of foundation models, 100 days of machine learning, Beethoven's 10th symphony, and more. Our latest focus series life on land (as part of our wider series on the UN sustainable development goals) was launched this month. We spoke to Lily Xu about her work in green security. Lily and her colleagues apply machine learning and game theory techniques to wildlife conservation.

As a result, you may send a photo to a deep neural network that has been trained to recognise dogs and cats and get an output that tells you whether the photo contains a dog or a cat. The network outputs the chance of the photo containing a dog or a cat (the two classes you trained it to identify) and the output sums to 100 per cent if the last network layer is a softmax layer. You get scores that you can interpret as probabilities of content belonging to each class, independently, when the last layer is a sigmoid-activated layer. The scores will not always add up to 100 per cent. Because its architecture outputs the entire image as being of a given class, a simple CNN can't duplicate the instances below.

Music artists, composers and producers today swim in massive amounts of musical notes to test the barriers of what melodies, harmonies and symphonies they can create and what works best with their songs. Although the advances in technology have significantly simplified and streamlined the process, it is still a long and challenging one for everyone involved in music creation. However, a technological revolution may be about to chance music creation as we know it. A team of computer scientists were able to use AI to complete the unfinished 10th symphony, originally created over 250 years ago by Ludwig Van Beethoven. This project has provoked interesting discussions, such as whether the now completed symphony is what Beethoven was originally trying to create, and also raised the important question -- what can Artificial Intelligence (AI) and Machine learning (ML) do for music production in the music entertainment industry? The team at Brainpool have been pondering on the answer to the latter, so we took the time to test a few of the various readily available AI music demos and reflected on how they could help transform the music industry.

Many developers have one common question – which of the framework is better? You might have got a fair idea by now of how each framework works and what are the best ways to use each of them. That said, there are many high traffic websites built on Flask as well, which perform well and are comparable with those built on Django. Quite a few basic concepts are the same in both Django and Flask, but Django is complicated and vast and requires a deep learning curve, so if one wants to get the'feel' of a web framework, start with Flask and then move on to Django for more sophisticated applications. Or you can master one and then switch when your job needs you to, rather than having to learn everything at once.

In the near future, we should see the value of AI-generated NFTs to expand beyond generative art into more generic NFT utility categories providing a natural vehicle for leveraging the latest deep learning techniques. An example of this value proposition can be seen in digital artists like Refik Anadol who are already experimenting with cutting edge deep learning methods for the creation of NFTs. Anadol's studio have been a pioneer in using techniques such as GANs, and even dabbling into quantum computing, trained models in hundreds of millions images and audio clips to create astonishing visuals. NFTs have been one of the recent delivery mechanisms explored by Anadol. In the near future, we should see the value of AI-generated NFTs to expand beyond generative art into more generic NFT utility categories providing a natural vehicle for leveraging the latest deep learning techniques.

Digital Art created via Convolutional Neural Networks and Deep Learning.

Digital Art created via Convolutional Neural Networks and Deep Learning.

In the prior post we explained how clustering of time series data works. In this post we're going to do a deep dive into the code itself. Everything will be written in python, but most libraries have an R version. We will try to stay relatively high level but the code will have some useful resources if you're looking for more. Without further ado, let's dive in.

Airborne magnetic data are commonly used to produce preliminary geological maps. Machine learning has the potential to partly fulfill this task rapidly and objectively, as geological mapping is comparable to a semantic segmentation problem. Because this method requires a high-quality dataset, we developed a data augmentation workflow that uses a 3D geological and magnetic susceptibility model as input. The workflow uses soft-constrained Multi-Point Statistics, to create many synthetic 3D geological models, and Sequential Gaussian Simulation algorithms, to populate the models with the appropriate magnetic distribution. Then, forward modeling is used to compute the airborne magnetic responses of the synthetic models, which are associated with their counterpart surficial lithologies. A Gated Shape Convolutional Neural Network algorithm was trained on a generated synthetic dataset to perform geological mapping of airborne magnetic data and detect lithological contacts. The algorithm also provides attention maps highlighting the structures at different scales, and clustering was applied to its high-level features to do a semi-supervised segmentation of the area. The validation conducted on a portion of the synthetic dataset and data from adjacent areas shows that the methodology is suitable to segment the surficial geology using airborne magnetic data. Especially, the clustering shows a good segmentation of the magnetic anomalies into a pertinent geological map. Moreover, the first attention map isolates the structures at low scales and shows a pertinent representation of the original data. Thus, our method can be used to produce preliminary geological maps of good quality and new representations of any area where a geological and petrophysical 3D model exists, or in areas sharing the same geological context, using airborne magnetic data only.

Feature selection techniques are essential for high-dimensional data analysis. In the last two decades, their popularity has been fuelled by the increasing availability of high-throughput biomolecular data where high-dimensionality is a common data property. Recent advances in biotechnologies enable global profiling of various molecular and cellular features at single-cell resolution, resulting in large-scale datasets with increased complexity. These technological developments have led to a resurgence in feature selection research and application in the single-cell field. Here, we revisit feature selection techniques and summarise recent developments. We review their versatile application to a range of single-cell data types including those generated from traditional cytometry and imaging technologies and the latest array of single-cell omics technologies. We highlight some of the challenges and future directions on which feature selection could have a significant impact. Finally, we consider the scalability and make general recommendations on the utility of each type of feature selection method. We hope this review serves as a reference point to stimulate future research and application of feature selection in the single-cell era.