Open rabernat opened 5 years ago
Next Tuesday I will give an overview of recurrent neural networks and empirical dynamic modeling to study nonlinear dynamical systems, drawing primarily from these two articles: https://www.ncbi.nlm.nih.gov/pubmed/23272922 and http://science.sciencemag.org/content/338/6106/496
Hi everyone. I have a pretty bad cold and have not managed to organize anything for next week. You're welcome to meet without me, but I think I won't make it.
I know many of you are planning to go to this lecture on Wednesday: https://events.r20.constantcontact.com/register/eventReg?oeidk=a07efl8m72m19af42c4&oseq=&c=&ch=
Let's discuss who will present next week (Oct. 23). Any volunteers?
I am happy to give an overview of building neural networks in python using Tensorflow/keras. I think it’s time we start getting our hands dirty with some code.
Last week there was some talk of some material already created for this? Maybe I misheard. If not, we could use a “classic” example such as predicting Boston housing prices using a plain old feed forward neural net or classifying hand written digits in the mnist dataset using a convolutional net.
Would anybody be interested in this?
I would be interested in this.
Luke, here is the material I mentioned:
https://github.com/raspstephan/ESS-Python-Tutorial
Note that weeks 6 and 7 of the UCI ESS python tutorial go through ML basics, using keras for image classification, and making a keras NN applied to a volcanic eruption data set. Already built as directed inquiry jupyter notebooks…
Cheers,
Mike.
Mike Pritchard Assistant Professor Dept of Earth System Sciences University of California, Irvine sites.uci.edu/pritchard
On Oct 15, 2018, at 11:30 AM, Luke Gloege notifications@github.com wrote:
I am happy to give an overview of building neural networks in python using Tensorflow/keras. I think it’s time we start getting our hands dirty with some code.
Last week there was some talk of some material already created for this? Maybe I misheard. If not, we could use a “classic” example such as predicting Boston housing prices using a plain old feed forward neural net or classifying hand written digits in the mnist dataset using a convolutional net.
Would anybody be interested in this?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/deepclimatenyc/seminar/issues/4#issuecomment-429964591, or mute the thread https://github.com/notifications/unsubscribe-auth/AZ91XC0QDwuclDr7t6c0hyxYqd-qXuBgks5ulNRDgaJpZM4XC6ee.
All,
Great! Thanks Mike for this material. I like the volcano eruption dataset. This is more interesting than the boring old Boston housing or MNIST datasets.
Unless others have some objection to this, Oct. 23 will be a "neural nets with Tensorflow/keras tutorial." I will put some material together and we will all learn how to build neural networks. In the next couple of days I will send out some material on how I setup a tensorflow environment in python.
Keras is a great way to learn neural networks since it has a very lego-block like structure, which makes it easy to whip-up a network is about 10 lines of code. It is also the API tensorflow has supported.
Here is how I envision the tutorial
Please let me know if you have suggestions on how to structure this tutorial. We only have an hour and thus need to be efficient with our time.
Everybody bring your laptop next week and let's build some networks! This is going to be fun!
-Luke
All,
Here are instructions on setting up a tensorflow-keras environment. Keras is now part of tensorflow, so no need to install it directly.
conda create --name keras
source activate keras
conda install -c conda-forge xarray
conda install -c conda-forge matplotlib
conda install -c conda-forge tensorflow
conda install -c conda-forge scikit-learn
conda install ipykernel
python -m ipykernel install --user --name keras --display-name "keras"
import xarray as xr
import numpy as np
import sklearn as sk
import tensorflow as tf
from tensorflow import keras
All,
@sridge will be joining us on Tuesday.
All,
Does anybody have a dongle for a circa 2014 macbook? If so, could I please use it today? Thanks!
Have you setup a keras environment yet? Don't worry, I've got your back. Everything is in a github repo and the example we will go through is accessible via Binder. We will go through a classic regression problem in machine learning: Predicting Boston housing prices. My goal is not to teach you everything about Keras, that is just not possible in an hour. The goals are this:
The tutorial repo and link to the binder notebook is here: https://github.com/lgloege/keras-tutorial
If this repo is deemed useful, somebody should incorporate it into the DL seminar repo.
Mike - I will add the material you mentioned to the repo so it can be accessed via binder. I did not have a chance to take a hard look at it and am not comfortable giving a tutorial on something I do not fully understand.
See you all at 3:00 today! This will be fun!
-Luke
Yes, I have all the dongles.
Thanks for organizing this Luke!
On Tue, Oct 23, 2018 at 10:58 AM Luke Gloege notifications@github.com wrote:
All,
Does anybody have a dongle for a circa 2014 macbook? If so, could I please use it today? Thanks!
Have you setup a keras environment yet? Don't worry, I've got your back. Everything is in a github repo and the example we will go through is accessible via Binder. We will go through a classic regression problem in machine learning: Predicting Boston housing prices. My goal is not to teach you everything about Keras, that is just not possible in an hour. The goals are this:
- Give a terse overview of what a neural network is doing
- Introduce how to use Keras to build a network
- Provide resources to build other types of networks (e.g. convolutional net)
The tutorial repo and link to the binder notebook is here: https://github.com/lgloege/keras-tutorial
If this repo is deemed useful, somebody should incorporate it into the DL seminar repo.
Mike - I will add the material you mentioned to the repo so it can be accessed via binder. I did not have a chance to take a hard look at it and am not comfortable giving a tutorial on something I do not fully understand.
See you all at 3:00 today! This will be fun!
-Luke
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/deepclimatenyc/seminar/issues/4#issuecomment-432280069, or mute the thread https://github.com/notifications/unsubscribe-auth/ABJFJiBWEiXJKLtusY3905NgkGgJtah0ks5uny6NgaJpZM4XC6ee .
Hi Luke,
Griffin Mooers (a UCI student) and I are huddled over a zoom terminal; will you be connecting us to today’s tutorial?
Thanks!
Mike.
Mike Pritchard Assistant Professor Dept of Earth System Sciences University of California, Irvine sites.uci.edu/pritchard
On Oct 23, 2018, at 7:58 AM, Luke Gloege notifications@github.com wrote:
All,
Does anybody have a dongle for a circa 2014 macbook? If so, could I please use it today? Thanks!
Have you setup a keras environment yet? Don't worry, I've got your back. Everything is in a github repo and the example we will go through is accessible via Binder. We will go through a classic regression problem in machine learning: Predicting Boston housing prices. My goal is not to teach you everything about Keras, that is just not possible in an hour. The goals are this:
Give a terse overview of what a neural network is doing Introduce how to use Keras to build a network Provide resources to build other types of networks (e.g. convolutional net) The tutorial repo and link to the binder notebook is here: https://github.com/lgloege/keras-tutorial https://github.com/lgloege/keras-tutorial If this repo is deemed useful, somebody should incorporate it into the DL seminar repo.
Mike - I will add the material you mentioned to the repo so it can be accessed via binder. I did not have a chance to take a hard look at it and am not comfortable giving a tutorial on something I do not fully understand.
See you all at 3:00 today! This will be fun!
-Luke
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/deepclimatenyc/seminar/issues/4#issuecomment-432280069, or mute the thread https://github.com/notifications/unsubscribe-auth/AZ91XGA7R6j25D774HrRCpZmgXOt4xdOks5uny6OgaJpZM4XC6ee.
There is no presentation scheduled for tomorrow. However, as described in #8, I have generated the qg-model datasets and placed them online. If anyone else wants to play around with these data, I'll be there at 3pm.
Hi all, I was wondering who might present at 3:00 pm tomorrow? I might present next week. Thanks a lot! Best, Yu Cheng
No one has volunteered. Unless someone volunteers or proposes a topic, we will not meet.
As no one has volunteered to present anything today, I am assuming we are not meeting.
I think Yu has volunteered for next week (Nov. 13)
Hi all,
I am going to present next Tuesday (Nov 20). Have a good day! Best, Yu Cheng
Is anyone planning to present tomorrow?
Sounds like no one wants to present today. I propose we attend this instead:
Tuesday, November 27, 2018 2:30PM - 4:00PM CESPR 412, Davis Auditorium
ABSTRACT: In August of 2018, the Texas Advanced Computing Center (TACC) at the University of Texas at Austin was selected as the sole awardee of the National Science Foundation’s “Towards a Leadership Class Computing Facility” solicitation. In this talk, I will describe the main components of the award: the Phase 1 system, “Frontera”, which will be the largest University-based supercomputer in the world when it comes online in 2019, the plans for for facility operations and scientific support for the next five years, and the plans to design a Phase 2 system in the mid-2020s to be the NSF Leadership system for the latter half of the decade, with capabilities 10x beyond Frontera. The talk will also cover the growing and shifting nature of the scientific workloads that require advanced capabilities, the technology shifts and challenges the community is currently facing, and the ways TACC has and is restructuring to face these challenges.
Bio: Dr. Stanzione is the Executive Director of the Texas Advanced Computing Center (TACC) at The University of Texas at Austin. A nationally recognized leader in high performance computing, Stanzione has served as deputy director since June 2009 and assumed the Executive Director post on July 1, 2014. He is the principal investigator (PI) for several leading projects including a multimillion-dollar National Science Foundation (NSF) grant to deploy and support TACC's Stampede supercomputer over four years. Stanzione is also the PI of TACC's Wrangler system, a supercomputer designed specifically for data-focused applications. He served for six years as the co-director of CyVerse, a large-scale NSF life sciences cyberinfrastructure in which TACC is a major partner. In addition, Stanzione was a co-principal investigator for TACC's Ranger and Lonestar supercomputers, large-scale NSF systems previously deployed at UT Austin. Stanzione previously served as the founding director of the Fulton High Performance Computing Initiative at Arizona State University and served as an American Association for the Advancement of Science Policy Fellow in the NSF's Division of Graduate Education. Stanzione received his bachelor's degree in electrical engineering and his master's degree and doctorate in computer engineering from Clemson University, where he later directed the supercomputing laboratory and served as an assistant research professor of electrical and computer engineering.
Hosted by Computing Systems for Data-Driven Science Center
Hello, I can present some of my recent results tomorrow. I would be presenting on "Estimating Surface Currents using Machine Learning"
We are in room 842A
Is anyone planning to join by zoom? If so just let me know and I'll start the meeting.
Hi everyone,
Here http://meetings.aps.org/Meeting/DFD18/SessionIndex2's a link to the APS DFD 2018 sessions, F32 and G32 are on Machine Learning & Data Driven Models. I’ve noticed that a lot of research is building from the first paper that we have discussed, Ling et al. 2016.
Cheers, Marco
On Dec 4, 2018, at 3:06 PM, Ryan Abernathey notifications@github.com wrote:
Is anyone planning to join by zoom? If so just let me know and I'll start the meeting.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/deepclimatenyc/seminar/issues/4#issuecomment-444239349, or mute the thread https://github.com/notifications/unsubscribe-auth/AEQ1WEoHTPG7_99cQuYixry9VxUPX3Ncks5u1tXZgaJpZM4XC6ee.
Here is the paper for tomorrow: https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.120.024102
May also discuss this a bit: https://aip.scitation.org/doi/abs/10.1063/1.5028373