Mirek Koldus

Mirek Koldus

@mir_k

Followers264
Following804

Digital Community Manager at DHL #techcommunity #innovation #machinelearning #ai #automation #iot #chatbots #vr #robotics

Prague
Joined on March 21, 2011

Here can be your content

Promote Yourself on Twipu

Promote Yourself on Twipu
Statistics

We looked inside some of the tweets by @mir_k and here's what we found interesting.

Inside 100 Tweets

Time between tweets:
2 days
Average replies
34
Average retweets
456
Average likes
1331
Tweets with photos
30 / 100
Tweets with videos
45 / 100
Tweets with links
10 / 100

Robots that learn to plug in cables, from images, with residual RL: https://arxiv.org/abs/1906.05841  Work from collaboration between RAIL and Siemens, led by Gerrit Schoettler & Ashvin Nair Some pretty nice cable insertion videos:

0
25
99
Question Answering models are data hungry. But what if there was no QA training data at all? Very excited to share "Unsupervised Question Answering by Cloze Translation" ACL 2019, with @LudovicDenoyer and @riedelcastro! https://arxiv.org/abs/1906.04980 

Question Answering models are data hungry. But what if there was no QA training data at all? Very excited to share "Unsupervised Question Answering by Cloze Translation" ACL 2019, with @LudovicDenoyer and @riedelcastro! https://arxiv.org/abs/1906.04980 

6
117
327
Check out a novel #ReinforcementLearning environment where agents aim to master the world’s most popular sport—football! The Google Research Football Environment includes benchmarks & progressive RL training scenarios, and is available in open source beta→http://goo.gle/2Mz8IqG 

Check out a novel #ReinforcementLearning environment where agents aim to master the world’s most popular sport—football! The Google Research Football Environment includes benchmarks & progressive RL training scenarios, and is available in open source beta→http://goo.gle/2Mz8IqG 

43
679
2K
Ever wanted to train a T-Rex to play soccer? Check out our new paper on learning composable motor skills with multiplicative compositional policies (MCP):
http://xbpeng.github.io/projects/MCP/ 

Big thanks to all my collaborators: @mmmbchang, Grace Zhang, @pabbeel, @svlevine

Ever wanted to train a T-Rex to play soccer? Check out our new paper on learning composable motor skills with multiplicative compositional policies (MCP): http://xbpeng.github.io/projects/MCP/  Big thanks to all my collaborators: @mmmbchang, Grace Zhang, @pabbeel, @svlevine

17
258
877
We're excited to announce a partnership with @Ford to develop last-mile delivery solutions that combine autonomous vehicles and legged robots. Full video at https://youtu.be/CUhuhIeQNos 

We're excited to announce a partnership with @Ford to develop last-mile delivery solutions that combine autonomous vehicles and legged robots. Full video at https://youtu.be/CUhuhIeQNos 

5
80
152
Watch these virtual, neural-net racing cars learn to drive. #machinelearning http://bit.ly/2Enkn5b 

Watch these virtual, neural-net racing cars learn to drive. #machinelearning http://bit.ly/2Enkn5b 

Want to play three quick games (some designed by humans, some by AI) and help me finish my thesis? Take part in this quick (~15 min) study! http://guzdial.com/experiment2019.html …

4
80
92

Trump bans US firms from doing business with foreign telcos

0
1
0
🐣 New Tutorial, open-source code & demo!

Building a SOTA Conversational AI with transfer learning & OpenAI GPT models

-Code/pretrained model from our NeurIPS 2018 ConvAI2 competition model, SOTA on automatic track
-Detailed Tutorial w. code
-Cool demo http://convai.huggingface.co 👇

🐣 New Tutorial, open-source code & demo! Building a SOTA Conversational AI with transfer learning & OpenAI GPT models -Code/pretrained model from our NeurIPS 2018 ConvAI2 competition model, SOTA on automatic track -Detailed Tutorial w. code -Cool demo http://convai.huggingface.co 👇

11
185
596
Straight from Google I/O, I'm happy to announce that @Arm and Google @TensorFlow are joining forces to accelerate TinyML. We will be merging uTensor, the first open source embedded inference project, with TensorFlow Lite Micro. https://os.mbed.com/blog/entry/uTensor-and-Tensor-Flow-Announcement/ … #tensorflow #tinyml #mbed

Straight from Google I/O, I'm happy to announce that @Arm and Google @TensorFlow are joining forces to accelerate TinyML. We will be merging uTensor, the first open source embedded inference project, with TensorFlow Lite Micro. https://os.mbed.com/blog/entry/uTensor-and-Tensor-Flow-Announcement/ … #tensorflow #tinyml #mbed

3
93
186
#io19 I wrote four short deep learning codelabs to get you started with neural networks. With TPUs you can now train and iterate in minutes. That's faster learning for our bio brains too! Head to the #codelabs area at I/O.

#io19 I wrote four short deep learning codelabs to get you started with neural networks. With TPUs you can now train and iterate in minutes. That's faster learning for our bio brains too! Head to the #codelabs area at I/O.

14
361
1K
Instead of computing the best racing line, I computed a huge grid of the "value" of being in any given position / angle, which lets a robot select the best action just by looking up all moves in a table. Bottom: table for the car's current heading. It recovers from any position.

Instead of computing the best racing line, I computed a huge grid of the "value" of being in any given position / angle, which lets a robot select the best action just by looking up all moves in a table. Bottom: table for the car's current heading. It recovers from any position.

This year, we have 3 papers at RSS, all on robotic RL: learning to walk, tool use w/ model-based RL, and learning from pixels w/o hand-designed rewards. All w/o using sim to train and on real robots
https://arxiv.org/abs/1812.11103 
https://arxiv.org/abs/1904.05538 
https://arxiv.org/abs/1904.07854 

This year, we have 3 papers at RSS, all on robotic RL: learning to walk, tool use w/ model-based RL, and learning from pixels w/o hand-designed rewards. All w/o using sim to train and on real robots https://arxiv.org/abs/1812.11103  https://arxiv.org/abs/1904.05538  https://arxiv.org/abs/1904.07854 

2
70
317
From Video Game to Real Robot: The Transfer between Action Spaces https://arxiv.org/abs/1905.00741 

From Video Game to Real Robot: The Transfer between Action Spaces https://arxiv.org/abs/1905.00741 

Our new paper, “Reinforcement Learning, Fast and Slow”, reviews recent techniques in deep RL that narrow the gap in learning speed between humans and agents, & demonstrate an interplay between fast and slow learning w/ parallels in animal/human cognition:

https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(19)30061-0 …

Our new paper, “Reinforcement Learning, Fast and Slow”, reviews recent techniques in deep RL that narrow the gap in learning speed between humans and agents, & demonstrate an interplay between fast and slow learning w/ parallels in animal/human cognition: https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(19)30061-0 …

13
734
2K

My recent talk at @MIT which covers a lot of our exciting new work applying AI to chess, StarCraft 2, and protein folding is now available on youtube here: https://www.youtube.com/watch?v=3N9phq_yZP0 … #AlphaZero #AlphaGo #AlphaStar

16
364
1K
I’m training a neural network to generate the plot for the next Marvel Avengers movie. I think it almost has it

I’m training a neural network to generate the plot for the next Marvel Avengers movie. I think it almost has it

11
16
108
I saw this video of Boston Dynamics' testing facility 5 days ago and I'm still thinking about it.

I saw this video of Boston Dynamics' testing facility 5 days ago and I'm still thinking about it.

532
6K
17K
Iterative Reinforcement Learning Based Design of Dynamic Locomotion Skills for Cassie https://arxiv.org/abs/1903.09537 

Iterative Reinforcement Learning Based Design of Dynamic Locomotion Skills for Cassie https://arxiv.org/abs/1903.09537 

Next Page