recentpopularlog in


« earlier   
Map Label Placement in Mapbox GL - Points of interest
An explanation by Ansis Brammanis of the extraordinary level of detail that goes into making labels work in Mapbox GL as maps are zoomed or rotated.
MapboxGL  labels  labelling  map  mapbox  AnsisBrammanis 
12 days ago by searchmeister
Adding Language to
Platform allows domain experts to produce high-quality labels for AI applications in minutes in a visual, interactive fashion. It combines the latest research in human perception, active learning, transfer from pre-trained nets, and noise-resilient training so that the labeler's time is used in the most productive way and the model learns from every aspect of the human interaction.
multi-modal  multi-class  multi-label  fastai  label  labels  labelling  deep-learning  nlp  architecture 
7 weeks ago by nharbour
ideonate/jupyter-innotater: Inline data annotator for Jupyter notebooks
Inline data annotator for Jupyter notebooks. Contribute to ideonate/jupyter-innotater development by creating an account on GitHub.
annnotate  data-annotate  annotation  label  labelling  deep-learning  juypter-notebook  dave-coates 
10 weeks ago by nharbour
Chemotherapy resistance and stemness in mitotically quiescent human breast cancer cells identified by fluorescent dye retention | SpringerLink
Metastatic recurrence in breast cancer is a major cause of mortality and often occurs many years after removal of the primary tumour. This process is driven by the reactivation of disseminated tumour cells that are characterised by mitotic quiescence and chemotherapeutic resistance. The ability to reliably isolate and characterise this cancer cell population is critical to enable development of novel therapeutic strategies for prevention of breast cancer recurrence. Here we describe the identification and characterisation of a sub-population of slow-cycling tumour cells in the MCF-7 and MDA-MB-231 human breast cancer cell lines based on their ability to retain the lipophilic fluorescent dye Vybrant® DiD for up to six passages in culture. Vybrant® DiD-retaining (DiD+) cells displayed significantly increased aldehyde dehydrogenase activity and exhibited significantly reduced sensitivity to chemotherapeutic agents compared to their rapidly dividing, Vybrant® DiD-negative (DiD−) counterparts. In addition, DiD+ cells were exclusively capable of initiating population re-growth following withdrawal of chemotherapy. The DiD+ population displayed only partial overlap with the CD44+CD24−/low cell surface protein marker signature widely used to identify breast cancer stem cells, but was enriched for CD44+CD24+ cells. Real-time qPCR profiling revealed differential expression of epithelial-to-mesenchymal transition and stemness genes between DiD+ and DiD− populations. This is the first demonstration that both MCF-7 and MDA-MB-231 human breast cancer lines contain a latent therapy-resistant population of slow-cycling cells capable of initiating population regrowth post-chemotherapy. Our data support that label-retaining cells can serve as a model for identification of molecular mechanisms driving tumour cell quiescence and de novo chemoresistance and that further characterisation of this prospective tumour-reinitiating population could yield novel therapeutic targets for elimination of the cells responsible for breast cancer recurrence.
stemness  labelling  vybrantDiD  MCF7  231  aldh  CD44  CD24  doxorubicin  paclitaxel  chemoresistance 
december 2019 by Segalllab
Human in the Loop: Deep Learning without Wasteful Labelling
The Oxford Applied and Theoretical Machine Learning Group (OATML) is a research group within the Department of Computer Science of the University of Oxford led by Prof Yarin Gal. We come from academia (Oxford, Cambridge, MILA, McGill, U of Amsterdam, U of Toronto, Yale, and others) and industry (Google, DeepMind, Twitter, Qualcomm, and startups). We follow pragmatic approaches to fundamental research in machine learning, and work with many practitioners in varied fields (medical, aut...
active-learning  activelearning  ai  and  annotations  applied  artificial  bayesian  bdl  deep  deep-learning  deeplearning  dl  fundamental  gal  gems  gprep  group  human  in  intelligence  labelling  learning  loop:  machine  ml  modelling  oatml  oxford  papers  pragmatic  probabilistic  resarch  research  the  theoretical  wasteful  without  yarin  yaringal  | 
december 2019 by linearpup
The viral selfie app ImageNet Roulette seemed fun – until it called me a racist slur • The Guardian
Julia Carrie Wong:
<p><a href="">ImageNet Roulette</a>, a project developed by the artificial intelligence researcher Kate Crawford and the artist Trevor Paglen…[aims] not to use technology to help us see ourselves, but to use ourselves to see technology for what it actually is.

The site’s algorithm was trained on photos of humans contained in ImageNet, a dataset described by Crawford as “one of the most significant training sets in the history of AI”. Created in 2007 by researchers at Stanford and Princeton, ImageNet includes more than 14m photographs, mostly of objects but also of humans, that have been classified and labeled by legions of workers on Amazon’s crowdsourcing labor site, Mechanical Turk.

If you upload your photo, ImageNet Roulette will use AI to identify any faces, then label them with one of the 2,833 subcategories of people that exist within ImageNet’s taxonomy. For many people, the exercise is fun. For me, it was disconcerting.

As a technology reporter, I’m regularly tasked with writing those scolding articles about why you should be careful which apps you trust, so I usually eschew viral face apps. But after a day of watching my fellow journalists upload their ImageNet Roulette selfies to Twitter with varying degrees of humor and chagrin about their labels (“weatherman”, “widower”, “pilot”, “adult male”), I decided to give it a whirl. That most of my fellow tech reporters are white didn’t strike me as relevant until later.

I don’t know exactly what I was expecting the machine to tell me about myself, but I wasn’t expecting what I got: a new version of my official Guardian headshot, labeled in neon green print: “gook, slant-eye”. Below the photo, my label was helpfully defined as “a disparaging term for an Asian person (especially for North Vietnamese soldiers in the Vietnam War)”.</p>

Which is also part of why diversity among journalists matters: because they can make a noise about it. If Wong had been just another user, her justifiable outrage would have been lost in the noise.
technology  face  labelling  imagenet 
september 2019 by charlesarthur
ideonate/jupyter-innotater: Inline data annotator for Jupyter notebooks
Inline data annotator for Jupyter notebooks. Contribute to ideonate/jupyter-innotater development by creating an account on GitHub.
jupyter  annotation  imageanalysis  deeplearning  extension  labelling  matplotlib 
september 2019 by wwwald

Copy this bookmark:

to read