This is a demo app showing off TensorFire’s ability to run the style-transfer neural network in your browser as fast as CPU TensorFlow on a desktop.
Continue reading “Fast Neural Style”
- Below is an extract from a 36-page report entitled “Technology and Innovation for the Future of Production: Accelerating Value Creation”, available for free here, and produced by the World Economic Forum.
- The extract below, about the future of AI, is figure 7 at page 13.
- This long report also discusses other interested topics and is peppered with many useful charts and illustrations.
- The following picture (figure 4 in the report) illustrating IoT is also interesting:
Below is an extract from a 36-page report entitled Technology and Innovation for the Future of Production: Accelerating Value Creation , available for free he…
Continue reading “Development of AI and its future state”
- For sheer volume of research on AI, if not quality, Chinese academics surpass their American peers; AI-related patent submissions in China almost tripled between 2010 and 2014 compared with the previous five years.
- No other country could generate such a volume of data to enable machines to learn patterns indicative of rare diseases, for example.
- A cyber-security law that came into force in June requires foreign firms to store data they collect on Chinese customers within the country’s borders; outsiders cannot use Chinese data to offer services to third parties.
- If it happens at all, the equivalent Chinese discussion about the limits of ethical AI research is far more opaque.
- AI techniques are perfect for finding patterns in the massive amounts of data that Chinese censors must handle in order to maintain a grip on the citizenry.
IMAGINE the perfect environment for developing artificial intelligence (AI). The ingredients would include masses of processing power, lots of computer-science boffins, a torrent of capital—and abundant data with which to train machines to recognise and respond to patterns.
Continue reading “Code redWhy China’s AI push is worrying”
- PredictN is a Prediction SaaS platform which automates all your business prediction requirements and helps improve efficiencies on your marketing and product development activities.
- Connect your data sources (GA 360, Adobe, BigQuery, CRM) and PredictN generates predictions directly to your marketing platforms (AdWords, DoubleClick, MailChimp).
Power your digital business actions using predictive machine learning
Continue reading “PredictN: Power your digital business actions using predictive machine learning – BetaList”
- The business came into its own in Paris in the 1960s when agencies began releasing “trend books”, collections of fabrics and design ideas.
- In response, forecasting agencies are making use of data collated from retailers’ IT systems and have added short-term predictions to their portfolio of services.
- EDITED, a competing service, provides “solid metrics” in fashion, claiming to use machine learning, an AI technique, in order to predict short-term sales trends.
- It releases a regular “Fashion Trends Report” based on the firm’s vast trove of search data.
- Whether AI will ever truly replace the woolly methods of fashion forecasting remains to be seen.
IN THE film “The Devil Wears Prada”, the character of Miranda Priestly, whose role is based on a feared Vogue editor, scolds her new assistant for not understanding fashion. Fashion, she tells her, is whatever a select group of designers and critics says it is.
Continue reading “AI la modeCan data predict fashion trends?”
- A convolutional layer operates over a local region of the input to that layer with the size of this local region usually specified directly.
- You can also compute the effective receptive field of a convolutional layer which is the size of the input region to the network that contributes to a layers’ activations.
- For example, if the first convolutional layer has a receptive field of 3×3 then it’s effective receptive field is also 3×3 since it operates directly on the input.
- However if the second layer of a convolutional network also has a 3×3 filter, then it’s (local) receptive field is 3×3, but it’s effective receptive field is 5×5.
A convolutional layer operates over a local region of the input to that layer with the size of this local region usually specified directly. You can also compute the effective receptive field of a convolutional layer which is the size of the input region to the network that contributes to a layers’ activations. For example, if the first convolutional layer has a receptive field of 3×3 then it’s effective receptive field is also 3×3 since it operates directly on the input.
Continue reading “Receptive Field Calculator”
- Subscribe & Review on iTunes, Soundcloud, Audioboom, Mixcloud & Acast, and join the discussion on Facebook and Twitter
In 2016 Klaus Schwab, founder and chairman of the World Economic Forum, wrote: “We stand on the brink of a technological revolution that will fundamentally alter the way we live, work, and relate to one another.”
- This fourth Industrial Revolution, he said, will fuse the physical, digital and biological worlds, and affect all corners of society – even challenging ideas of what it means to be human.
- Andrew McAfee and Erik Brynjolfsson of the the Massachusetts Institute of Technology think not.
- And in their latest book Machine, Platform, Crowd, they tell us why.
- Joining Ian Sample in the studio, Andrew and Erik lay out their blueprint for the future of the workplace, including the role big data will play, how some processes involving decision-making could be automated, and how minds and machines can come together to cancel out each other’s errors.
Ian Sample sits down with Andrew McAfee and Erik Brynjolfsson to discuss the future of the workplace and the role artificial intelligence will play
Continue reading “Minds and machines: can we work together in the digital age?”
- Read the full ABC article and watch the video interview to learn more about Tanmay and his work in the field of AI.
- The Australian Broadcasting Corporation (ABC) recently profiled 13-year-old Canadian tech prodigy Tanmay Bakshi who started using computers at age five, launched his first app at age nine, and has been working with IBM’s AI and cognitive APIs for a couple of years now.
- ABC notes: “the Canadian teen has become a global force in programming and commands more than 20,000 subscribers on his YouTube channel that teaches computer coding.”
- He is currently in Australia for the IBM Watson Summit, which brings together experts in artificial intelligence to discuss how the technology can help people and businesses in the future.
- You can also watch Tanmay’s video, “IBM Watson, Machine Learning: How to use the “Retrieve and Rank” service in IBM Bluemix”, one of 80 tutorials he has created and made available on the “Tanmay Teaches” YouTube channel.
A profile of 13-year-old Canadian tech prodigy Tanmay Bakshi who started using computers at age five and works with IBM’s AI and cognitive APIs.
Continue reading “Meet the 13-year-old prodigy taking IBM and artificial intelligence by storm”
- A key learning, is that the way in which these SVM’s are structured can actually have a significant impact on how much training data has to be applied, for example, a naive approach would have been as follows:
This approach requires that for every additional sub-category, two new SVM’s be trained – for example, the addition of a new class for ‘Swimwear’ would require an additional SVM under Men’s and Women’s – not to mention the potential complexity of adding a ‘Unisex’ class at the top level.
- We were able to avoid a great deal of labelling& training work, by flattening our data structures into many sub-trees like so:
By decoupling our classification structure from the final hierarchy, it is possible to generate the final classification by traversing the SVM hierarchy with each document, and interrogating the results with simple set-based logic such as:
Mens Slim-fit jeans = (Mens and Jeans and Slim Fit) and not Womens
This approach vastly reduces the number of SVM’s required to classify documents, as the resultant sets can be intersected to represent the final classification.
- For example – adding a top-level ‘Children’s’ class – would immediately allow the creation of an entire dimension of new Children’s categories (children’s jeans, shirts, underwear, etc), with minimal additional training data (Only one additional SVM):
Because of the structure we chose, one key insight that we were able to leverage, was that of re-using training data, via linked data relationships.
- For example, given some basic domain knowledge of the categories – we know for certain that ‘Washing machines’ can never be ‘Carpet cleaners’
By adding the ability to link ‘Exclude data’, we can heavily bolster the amount ‘Negative’ training examples for the ‘Washing machines’ SVM by adding to it the ‘Positive’ training data from ‘Carpet cleaners’ SVM.
- This approach has a nice uptick, in that whenever the need arises to add some additional training data to improve the ‘Carpet Cleaners’ SVM – it inadvertently improves the ‘Washing machines’ class, via linked negative data.
In many cases, getting enough well-labelled training data is a huge hurdle for developing accurate prediction systems. Here is an innovative approach which uses SVM to get the most from training data.
Continue reading “How to squeeze the most from your training data”
- Tesla CEO Elon Musk and Facebook CEO Mark Zuckerberg are waging a public debate over the merits of AI.
- Musk has said in the past that AI could be potentially very damaging to humans, and Zuckerberg recently called such doomsday predictions “irresponsible.”
- Musk responded on Twitter, calling Zuckerberg’s understanding of AI “limited.”
Elon Musk and Mark Zuckerberg are currently embroiled in a public debate over the future of AI.
Continue reading “Elon Musk and Mark Zuckerberg disagree about the future of AI”