jm + labelling   3

Google Hired Gig Economy Workers for Project Maven
Other tech giants are reportedly interested in engaging the military as it continues to deploy artificial intelligence technology. Much larger machine-learning projects may require vastly new engagement from gig economy workers, who may unknowingly engage in the work.

“Workers absolutely should have the right to know what they are working on, and especially when moral or politically controversial activities are involved,” said Juliet Schor, a sociology professor at Boston College, in an email to The Intercept. “It’s a basic dimension of democracy, which should not stop at either the factory or the platform ‘door.’ For too long, the country has tolerated erosion of basic civil rights in the workplace, as corporations assume ever-more control over their workforces. It’s time to win them back.”
google  project-maven  ai  training  labelling  work  ethics  military 
19 days ago by jm
Where your "full Irish" really comes from
This is really disappointing; many meats labelled as "Irish" are anything but. The only trustworthy mark is the Bord Bia "Origin Ireland" stamp -- I'll be avoiding any products without this in future.
Under European labelling law, country of origin is mandatory for beef, fish, olive oil, honey and fresh fruit and vegetables. Next month the EU will make it law to specify country of origin for the meat of pigs, chicken, sheep and goats, with a lead-in time of anywhere up to three years for food companies to comply.
The pork rule, however, will only apply to fresh pork and not to processed meat, so consumers still won’t get a country-of-origin label on rashers, sausages or ham. In the meantime, the Bord Bia Origin-Ireland stamp is a guarantee that your Irish breakfast ingredients are indeed Irish.
bord-bia  labelling  eu  country-of-origin  meat  pork  food  quality 
november 2013 by jm
_Building High-level Features Using Large Scale Unsupervised Learning_ [paper, PDF]
"We consider the problem of building highlevel, class-specific feature detectors from only unlabeled data. For example, is it possible to learn a face detector using only unlabeled images using unlabeled images? To answer this, we train a 9-layered locally connected sparse autoencoder with pooling and local contrast normalization on a large dataset of images (the model has 1 billion connections, the dataset has 10 million 200x200 pixel images downloaded from the Internet). We train this network using model parallelism and asynchronous SGD on a cluster with 1,000 machines (16,000 cores) for three days. Contrary to what appears to be a widely-held intuition, our experimental results reveal that it is possible to train a face detector without having to label images as containing a face or not. Control experiments show that this feature detector is robust not only to translation but also to scaling and out-of-plane rotation. We also find that the same network is sensitive to other high-level concepts such as cat faces and human bodies. Starting with these learned features, we trained our network to obtain 15.8% accuracy in recognizing 20,000 object categories from ImageNet, a leap of 70% relative improvement over the previous state-of-the-art."
algorithms  machine-learning  neural-networks  sgd  labelling  training  unlabelled-learning  google  research  papers  pdf 
june 2012 by jm

Copy this bookmark:



description:


tags: