f(AI)nder View
February 2019
f(AI)nder View - a monthly digest in the field of
Artificial Intelligence, Machine Learning, Deep Learning
exploring the various levels and faces of AI: from basic to sophisticated ones,
from algorithms and technologies to business applications

Quadrant - partner of the February issue
Empowering Data Professionals
In the 21st century economy, data is key to success. But data can be chaotic, confusing, and in some cases, fake. Quadrant provides access to data that is relevant and trustworthy, mapping disparate inputs into reliable, intelligible information that addresses uncertainty when it comes to using data.
The platform uses blockchain technology - the best technical solution currently available for this purpose - to ensure the data is authentic, trustworthy, and useable.
With Quadrant, organisations and individuals can now have full trust in their data and use it to build targeted solutions and allocate resources efficiently to meet the requirements of their customers, citizens, and colleagues.
AI ecosystem
AI industry analytics, trends, foresights, insights, challenges, regulations, etc.
CB Insights' third annual cohort of AI 100 startups is a list of 100 of the most promising private companies providing hardware and data infrastructure for AI applications, optimizing machine learning workflows, and applying AI across a variety of major industries. The AI 100 companies are disrupting 12 core sectors, including healthcare, telecommunications, semiconductors, government, retail, and finance, as well as the broader enterprise tech stack.
Read more
The report reveals trends in patenting of artificial intelligence (AI) innovations, the top players in AI from industry and academia, and the geographical distribution of AI-related patent protection and scientific publications. Since artificial intelligence emerged in the 1950s, innovators and researchers have filed applications for nearly 340,000 AI-related inventions and published over 1.6 million scientific publications.
Read more
In the Accenture Technology Vision 2019 survey of more than 6,600 business and IT executives. New technologies are a catalyst for change. The next set of technologies every company will need to master? DARQ: distributed ledger technology (DLT), artificial intelligence (AI), extended reality (XR) and quantum computing. DARQ technologies will be the next source of differentiation and disruptive technology.
Read more
Europe's average digital gap with the world's leaders is now being compounded by an emerging gap in artificial intelligence. Only two European companies are in the worldwide digital top 30, and Europe is home to only 10 percent of the world's digital unicorns. Europe has about 25 percent of AI startups, in line with its size in the world economy, but its early-stage investment in AI lags behind that of the United States and China.
Read more
The main cluster cities (Montreal, Toronto, Vancouver) have seen pillar companies (e.g. Google, Uber, Facebook) setting up research groups, adding to already very active research communities. The continuous support to public research labs also sheds light on why the Canadian AI Ecosystem has one of the biggest talent pools in the world. When compared to other national ecosystems, Canada hosts the third largest number of AI experts.
Read more
AI companies are defined as those that use machine learning as a core differentiator, sell AI software, or build AI chips. There are 32 companies in the AI Unicorn Club by now. AI startups valued at $1B+ nearly doubled in 2018. A total of 17 AI startups reached $1B+ valuations in 2018, up from 9 the previous year. From banking fraud prevention to handheld ultrasounds to recruiting software, here's the newest crop of AI unicorns.
Read more
Applications
How do you visualize the structure underlying artistic creation? One model is, of course, the modern museum. But there is another model—made possible by artificial intelligence—that may provide an outlook difficult to translate in brick-and-mortar institutions. Generist Maps makes use of technology called generative adversarial networks (or GANs), which are a type of neural network.In Generist Maps's case, the GAN powering the prototype looks at the areas of art represented within The Met collection and infers the "latent space" between them by simulating intermediary images.
Read more
With the help of deep learning techniques, paleoanthropologists find evidence of long-lost branches on the human family tree. The new deep learning method is an attempt to explain levels of gene flow that are too small for the usual statistical approaches, and by offering a far more vast and complicated range of models to do so. This use of deep learning can uncover ghosts we didn't even suspect. For one, there's no reason to think that Neanderthals, Denisovans and modern humans were the only three populations in the picture.
Read more
Each time you refresh the site, the network will generate a new facial image from scratch. The algorithm behind it is trained on a huge dataset of real images, then uses a type of neural network known as GAN to fabricate new examples.The algorithm, named StyleGAN, was made open source recently and has proven to be incredibly flexible. Although this version of the model is trained to generate human faces, it can, in theory, mimic any source. Researchers are already experimenting with other targets, including anime characters, fonts, and graffiti.
Read more
A chatbot builder, also known as a chatbot development platform, is an application where a user creates a chatbot for the web, app or popular messaging platform.
Some platforms can allow for deployment of chatbots to voice applications such as Google Home and Amazon Alexa while others provide a low-code interface with drag-and-drop and visual editor.
There are even chatbot builders that leverage machine learning to continuously learn every interaction so that they can deliver an optimized response.
Read more
There are many ways artificial intelligence (AI) and machine learning can make our world more productive and effective. There are even breweries that are using AI to enhance beer production. Since brewing beer is an art and a science, artificial intelligence offers a powerful helping hand in the latter. IntelligentX has the distinction of creating the world's first beer that used AI algorithms and machine learning to help adjust its recipe. IntelligentX creates four different varieties of beer—Black AI, Golden AI, Pale AI, and Amber AI.
Read more
RePaint uses a combination of 3-D printing and deep learning to authentically recreate favorite paintings — regardless of different lighting conditions or placement.
RePaint could be used to remake artwork for a home, protect originals from wear and tear in museums, or even help companies create prints and postcards of historical pieces. RePaint was more than four times more accurate than state-of-the-art physical models at creating the exact color shades for different artworks.
Read more
Introductions
Strategy for DL troubleshooting: Start simple - Choose the simplest model & data possible; Implement & debug - Once model runs, overfit a single batch & reproduce a known result; Evaluate - Apply the bias-variance decomposition to decide what to do next; Tune hypeparams - Use coarse-to-fine random searches; Improve model/data - Make your model bigger if you underfit, add data or regularize if you overfit.
Read more
OpenAI team has trained a large-scale unsupervised language model which generates coherent paragraphs of text, achieves state-of-the-art performance on many language modeling benchmarks, and performs rudimentary reading comprehension, machine translation, question answering, and summarization — all without task-specific training.
Read more
Within just the past 1 or 2 years the field has seen a lot of excitement over a potentially powerful approach known as the graph network. These are deep-learning systems that have an innate bias toward representing things as objects and relations.
The graph-network approach has already demonstrated rapid learning and human-level mastery of a variety of applications, including complex video games.
Read more
Generative Adversarial Networks are a powerful class of neural networks with remarkable applications. They essentially consist of a system of two neural networks — the Generator and the Discriminator — dueling each other. GANs have a plethora of applications, as they can learn to mimic data distributions of almost any kind. Popularly, GANs are used for removing artefacts, super resolution, pose transfer, and literally any kind of image translation.
Read more
Can we build systems that can learn faster, and with less data? "Meta-learning'', one of the most exciting ML research topics right now, addresses this problem by optimizing a model not just for the ability to "predict well'', but also the ability to "learn well''.
An introduction to "what is meta-learning" and a tutorial on implementing Model-Agnostic Meta-Learning (MAML) in about 50 lines of Python code, using Google's JAX library. A self-contained Jupyter notebook, reproducing the tutorial, is here.
Read more
"In short, PlaNet learns a dynamics model given image inputs and efficiently plans with it to gather new experience. In contrast to previous methods that plan over images, we rely on a compact sequence of hidden or latent states. This is called a latent dynamics model: instead of directly predicting from one image to the next image, we predict the latent state forward. The image and reward at each step is then generated from the corresponding latent state."
Read more
Toolbox
SC-FEGAN is a novel image editing system that generates images as the user provides free-form mask, sketch and color as an input. The system consists of a end-to-end trainable convolutional network. Architecture SC-FEGAN is well suited to generate high quality synthetic image using intuitive user inputs.
Read more
BioBERT (Bidirectional Encoder Representations from Transformers for Biomedical Text Mining) is a domain specific language representation model pre-trained on large-scale biomedical corpora. Based on the BERT architecture, BioBERT effectively transfers the knowledge from a large amount of biomedical texts to biomedical text mining models with minimal task-specific architecture modifications.
Read more
LASER (Language-Agnostic SEntence Representations) toolkit now works with more than 90 languages, written in 28 different alphabets. LASER achieves these results by embedding all languages jointly in a single shared space. LASER sets a new state of the art on zero-shot cross-lingual natural language inference accuracy for 13 of the 14 languages in the XNLI corpus.
Read more
A quick overview of the most interesting publications from 2018 regarding GANs (part 1 & part 2): GAN Dissection: Visualizing and Understanding Generative Adversarial Networks; A Style-Based Generator Architecture for Generative Adversarial Networks; Evolutionary Generative Adversarial Networks; Large Scale GAN Training for High Fidelity Natural Image Synthesis and others.
Read more
Any of the data structures can be formalized to graphs. For instance an image can be seen as grid of pixel, text a sequence of words… DGL is a Python package that interfaces between existing tensor libraries and data being expressed as graphs. It makes implementing graph neural networks (including Graph Convolution Networks, TreeLSTM, and many others) easy while maintaining high computation efficiency.
Read more
The review covers a wide range of open source software for quantum computing, covering all stages of the quantum toolchain from quantum hardware interfaces through quantum compilers to implementations of quantum algorithms, as well as all quantum computing paradigms, including quantum annealing, and discrete and continuous-variable gate-model quantum computing.
Read more
AI hardware
The Center will host research and development, emulation, prototyping, testing, and simulation activities for new AI cores specially designed for training and deploying advanced AI models, including a test bed in which members can demonstrate Center innovations in real-world applications. A key area of research and development will be systems that meet the demands of deep learning inference and training processes.
Read more
AI-related semiconductors will see growth of about 18 percent annually over the next few years—five times greater than the rate for semiconductors used in non-AI applications. By 2025, AI-related semiconductors could account for almost 20 percent of all demand, which would translate into about $67 billion in revenue. Opportunities will emerge at both data centers and the edge.
Read more
Neuromorphic engineering points to the possibility, if not yet probability, of a massive leap forward in performance, by way of a radical alteration of what it means to infer information from data. Like quantum computing, it relies upon a force of nature we don't yet comprehend. A truly neuromorphic device, its practitioners explain, would include components that are physically self-assembling.
Read more
Having a fast GPU is a very important aspect when one begins to learn deep learning as this allows for rapid gain in practical experience which is key to building the expertise with which you will be able to apply deep learning to new problems. Without this rapid feedback, it just takes too much time to learn from one's mistakes and it can be discouraging and frustrating to go on with deep learning.
Read more
Xnor.ai (Xnor) has announced about its new innovation, i.e. a standalone battery-free solar-powered AI technology. The device will continuously run for 32 years on a coin cell battery; detecting things every second. The new technology allows AI to run constantly without any charge or power on edge devices. Xnor's solutions identify visual objects; even people. Now the models can be powered on a simple solar cell instead of powerful GPUs which consumed more energy.
Read more
LeCun says the demand for DL-specific hardware will likely only increase. New architectural concepts such as dynamic networks, associative-memory structures, and sparse activations will affect the type of hardware architecture that will be required in the future. Watch the video to see LeCun discuss the hardware challenges the industry must address to create dramatically more effective and efficient AI systems.
Read more
AI competitions
AI Supernova Challenge is open to ALL startups. $100,000 Prize Pool. Four winners will be selected to win the prizes.
Deadline to apply: April 1st, 2019.
Read more
In this competition, you will address when the earthquake will take place.
$50,000 Prize Money.
Entry deadline: May 27, 2019.
Read more
IJCAI-19 Alibaba Adversarial AI Challenge
The competition focus on image classification tasks, and includes model attacks and model defenses. Competition Schedule: February 15 - May 31, 2018.
Read more
Events
14-17 March, 2019. Singapore.
The tracks and sessions about Cloud, DevOps, AI, Machine Learning, Blockchain, Open Hardware, Science, Web, Mobile and much more.
Read more
19-21 March, 2019. Boston, USA.
Topic Coverage: Quantum Computing Hardware And Software, Quantum Computing Applications, Quantum Encryption And Quantum Networking, Quantum Components.
Read more
Future of AI 2019
26-27 March, 2019. Lago, Israel.
Meet the leaders of AI: hundreds of startups, investors, and senior decision makers in the world's leading companies creating AI technology, products and services.
Read more
Follow us
f(AI)nder View - a monthly digest in the field of
Artificial Intelligence, Machine Learning, Deep Learning
exploring the various levels and faces of AI:
from basic to sophisticated ones,
from algorithms and technologies to business applications

f(AI)nder View is prepared by f(AI)nder Lab
Twitter
Made on
Tilda