Discover and read the best of Twitter Threads about #neuralnetworks

Most recents (11)

Please help us welcome our next curator Darryl Takudzwa Griffiths. @BlaqNinja completed his Bachelors Degree in Computer Engineering at DUT, graduated in 2011. Due to struggling to find suitable employment he went on to study multiple certificates from bodies such as Microsoft.
He has certificates in N+ (Computer Networking), A+ (Computer Technician & Technical Support), Certified Ethical Hacking V7 (CEH v7), Offensive Security Certified Professional (OSCP). Sadly even with these, he could not secure his desired post so in 2016 he moved to USA.
Darryl was able to secure a job in a corporation that owns casinos as a system analyst & security architect. Within the same year he embarked on a Masters degree in Robotics & Artificial Intelligence Engineering. In 2017 he resigned from his post and started his own company...
Read 99 tweets
Es un orgullo para la Comunidad de Desarrolladores de Argentina poder acompãnar iniciativas como el #ConnectDay junto a estas empresas @plataforma5la, @distillerylatam, @revistasg y @clarikagroup 💪
¡Hoy es el #ConnectDay! Desde CoDeAr estamos felices de poder acompañar a @wtmriodelaplata, @GDGCordobaARG, @gdgriodelaplata en este día de charlas y de compartir conocimiento en comunidad. Podés sumarte a la transmisión en vivo desde acá:
Comienza la primer charla sobre #DataScience y #Economía, en el contexto de las #transdisciplinas.
Read 118 tweets
PathME unsupervised #multiomics
1) genes space → pathways space
2) for each pathway: collapse pathways from multiple omics to one per patient
3) sparse NMF biclustering

✓compared against SNF and iCluster
✓TCGA x 4
✓hyperparameters
✓source code
✓5-fold CV

#SundayMultiOmics
Worth noting:
- authors use sNMF consensus from 500 runs (cophenetic correlation + permutation testing to choose # of clusters)
- the autoencoders are denoising
- worth praise is the effort into interpretability (of both features/omics & clinical associations) - see supplement!
Read 9 tweets
A "worrying analysis":

"18 [#deeplearning] algorithms ... presented at top-level research conferences ... Only 7 of them could be reproduced w/ reasonable effort ... 6 of them can often be outperformed w/ comparably simple heuristic methods."

Paper:
lnkd.in/dTaGCTv

#AI
[Updates worth tweeting]

2/
There is much concern about #reproducibility issues and flawed scientific practices in the #ML community in particular & #academia in general.

Both the issues and the concerns are not new.

Isn't it time to put an end to them?
3/
There are several works that have exposed these and similar problems along the years.

👏👏 again to @Maurizio_fd et al. for sharing their paper and addressing #DL algorithms for recommended systems (1st tweet from this thread).

But there is more, unfortunately:
Read 18 tweets
Thanks to @turo I'm renting a #TeslaModel3 during my visit to Silicon Valley - certainly a different experience than traditional #carrental agencies #future #zeroemission #electriccars
Gone to Tesla's HQ in Fremont, California to try the supercharging there. It's supposed to be much faster than before.
I had hoped the Model 3 would charge at the new rate but it didn't happen
Read 51 tweets
Our new #Corl18 paper:

How curiosity-driven autonomous goal setting enables to discover
independantly controllable features of environments

pdf: arxiv.org/abs/1807.01521

Blog: openlab-flowers.inria.fr/t/discovery-of…

Colab: colab.research.google.com/drive/176q8pns…

#machinelearning #AI #NeuralNetworks
Imagine a robot perceiving a scene through low-level pixels.

It has no prior knowledge of its body (it only knows it can send numerical
numbers as motor commands that produce unknown movements of unknow body parts)
It also does not know that the scene is composed of "objects". Yet, around it there are several objects: some of them can be controlled, others not (distractors)

How could it discover and represent entities, and find out which ones are controllable (and learn to control them)?
Read 24 tweets
Just about to kick off the keynotes at #ODSCWest - also to come today: job fair, hands-on workshops & networking. First speaker: @karpathy
Key element of #AI is that neural networks are code. It's just written by machines - #Code2point). Still need to be maintained, develop tests, refactoring and organization. #ODSC
Need to set up unit tests for #AI neural networks. How to meet tests, tune dataset, tune model, tune optimization. #ODSCWest
Read 20 tweets
Neural ensembles are groups of co-active neurons that may be triggered spontaneously, by sensory stimuli or behavior. Such ensembles are therefore likely to constitute the building blocks of brain function, but little is known about their structure, organization and dynamics.
Studying neural ensembles relies on our ability to identify them. Unlike existing methods, @Giovann82176354 provides a model based approach that uses Bayesian inference techniques to obtain probabilistic estimates of the number, composition and dynamics of neural ensembles.
By applying this method my 2-photon volumetric recordings of spontaneous activity in the optic tectum we recover the structure of multiple spatially compact ensembles spanning multiple imaging planes. (note: this visualization was controlled by a Nintendo 64!)
Read 7 tweets
1. Let's talk about trinary ( #ternary ) and #IOTA . (Long rant, maybe better suited for a blog entry. Oh well, whatever.)
2. In my many discussions with many different people about #IOTA , ternary is I think the most controversial aspect of the tech. Many people consider it genuine lunacy, a sign that #IOTA is just a scam cryptocurrency.
3. And I'll be honest with you: when I first encountered ternary in #IOTA , I went "What the f*** is this sh**? Numerology?"
Read 21 tweets
OK people. I'm live tweeting @neo4j's #Chicago #GraphTour event today.
We are currently in the keynote. The speaker is talking about how popular Neo4j is
Keynote now telling us about all the new Neo4j features. Examples: location filter, including 3d. Auto cache reheating. I'm interested to know if auto cache reheating is working with query patterns or meant to replace them or what
Read 64 tweets
How many random seeds are needed to compare #DeepRL algorithms?

Our new tutorial to address this key issue of #reproducibility in #reinforcementlearning

PDF: arxiv.org/pdf/1806.08295…

Code: github.com/flowersteam/rl…

Blog: openlab-flowers.inria.fr/t/how-many-ran…

#machinelearning #neuralnetworks
Algo1 and Algo2 are two famous #DeepRL algorithms, here tested
on the Half-Cheetah #opengym benchmark.

Many papers in the litterature compare using 4-5 random seeds,
like on this graph which suggests that Algo1 is best.

Is this really the case? Image
However, more robust statistical tests show there are no differences.

For a very good reason: Algo1 and Algo2 are both the same @OpenAI baseline
implementation of DDPG, same parameters!

This is what is called a "Type I error" in statistics.
Read 11 tweets

Related hashtags

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3.00/month or $30.00/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!