Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Using our brain simulator, we’ve trained a reinforcement learning agent to maximize bits per second. Here is the RL policy converting brain data to cursor control in simulation:

Neuralink

1,709,999 subscribers

18,410 views • 1 year ago •via X (Twitter)

Science & Technology Education

Anya Rossi• Live Now

Private livecam show

12 Comments

Neuralink1 year ago

🧵Deep learning simulation of brain activity

Neuralink1 year ago

Just like self-driving cars, BCIs are control systems. High-fidelity simulations of motor cortex activity can speed up evaluation of neural decoders and enable optimization methods like reinforcement learning.

Neuralink1 year ago

High performance in simulation is exciting, but what happens when we deploy this neural decoding policy in the real world? If our simulation is accurate, users should achieve similar performance in real-world applications.

Neuralink1 year ago

We’re in the early stages of generative brain modeling, but Pager can already use decoders trained in simulation to acquire targets in the real world:

Neuralink1 year ago

holds the BPS world record of 9.5 bits per second (bps). The challenge now: can this approach scale to 10 bps and beyond for our participants?

Neuralink1 year ago

If you want to answer this question, apply here:

HUDI1 year ago

🚨 AI is coming to HUDI! 🚀 We’re working on bringing you a private AI to 💬 “talk with your data” for valuable insights. 🐸 With DataMask, the first self-custody data wallet, you have full control. Manage, share, and monetize your data with daily grants, quests, and surveys on HUDI dapp. Reclaim your data’s value and start earning soon with HUDI! 🌟 #AI #DataPrivacy #DataMonetization #HUDI #DataMask #SelfCustody #DailyGrant #TechForGood #ReclaimYourData @Jason thanks for the meme base 🤣❤️ have a look at our adventure updates!

Ninus1 year ago

Is it possible we take a look at the EEG associated with this?

High Jack1 year ago

Great advancement! High-fidelity simulations + RL can revolutionize BCIs. Excited to see this evolve!

Jaitan Martini1 year ago

Wow

Heather Mascoe1 year ago

EEG vs. EKG Brain vs. Heart 🧠 vs. ♥️

Nettie Travelling1 year ago

👍

Related Videos

TWIST: a real-time teleoperation system for humanoid robots to mimic whole-body motions. Reference motion data is generated by retargeting human motion-capture data to the robot. Then, the controller is trained in simulation using reinforcement learning and behavior cloning.

TWIST: a real-time teleoperation system for humanoid robots to mimic whole-body motions. Reference motion data is generated by retargeting human motion-capture data to the robot. Then, the controller is trained in simulation using reinforcement learning and behavior cloning.

The Humanoid Hub

46,715 views • 1 year ago

Simulation provides the ability to unlock autonomy at scale. Our new blog shows how multi-agent reinforcement learning allows us to populate our Wayve Infinity Simulator with the emergent behaviour critical to represent dynamic urban cities in the real world.

Simulation provides the ability to unlock autonomy at scale. Our new blog shows how multi-agent reinforcement learning allows us to populate our Wayve Infinity Simulator with the emergent behaviour critical to represent dynamic urban cities in the real world.

Alex Kendall

66,919 views • 3 years ago

i trained a reinforcement learning agent to play diablo 1 in a few hours with pufferlib

i trained a reinforcement learning agent to play diablo 1 in a few hours with pufferlib

Dan Advantage

41,895 views • 6 months ago

Our course recommendation of the day is “Post-training of LLMs, ” where you’ll learn how to customize pre-trained language models using Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Online Reinforcement Learning (RL). You'll learn when to use each method, how to curate training data, and implement them in code to shape model behavior effectively. Enroll here:

Our course recommendation of the day is “Post-training of LLMs, ” where you’ll learn how to customize pre-trained language models using Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Online Reinforcement Learning (RL). You'll learn when to use each method, how to curate training data, and implement them in code to shape model behavior effectively. Enroll here:

DeepLearning.AI

29,369 views • 8 months ago

I set the world record for cursor control with a brain computer interface using my Neuralink brain implant! #Neuralink #bci #fyp

I set the world record for cursor control with a brain computer interface using my Neuralink brain implant! #Neuralink #bci #fyp

Jake Schneider

11,932 views • 10 months ago

New research reveals the speed of human thought: 10 bits per second. Why does the brain think so slowly when our senses process data 100 million times faster? A paradox that reshapes our understanding of cognition.

New research reveals the speed of human thought: 10 bits per second. Why does the brain think so slowly when our senses process data 100 million times faster? A paradox that reshapes our understanding of cognition.

Neuroscience News

31,736 views • 1 year ago

The needle grasps and inserts each electrode thread of our implant into the brain. We've built the technology to produce these needles in-house, using lasers to mill features invisible to the naked eye.

The needle grasps and inserts each electrode thread of our implant into the brain. We've built the technology to produce these needles in-house, using lasers to mill features invisible to the naked eye.

Neuralink

206,169 views • 1 month ago

Haven't been to a conference in a while, really excited to be at #NeurIPS2024! I'll be helping present 4 of our group's recent papers: 1. Overcoming the Sim-to-Real Gap: Leveraging Simulation to Learn to Explore for Real-World RL 2. Distributional Successor Features Enable Zero-Shot Policy Optimization 3. Learning to Cooperate with Humans using Generative Agents 4. Personalizing Reinforcement Learning from Human Feedback with Variational Preference Learning Find more details on each paper and where to find us in this thread (1/6)

Haven't been to a conference in a while, really excited to be at #NeurIPS2024! I'll be helping present 4 of our group's recent papers: 1. Overcoming the Sim-to-Real Gap: Leveraging Simulation to Learn to Explore for Real-World RL 2. Distributional Successor Features Enable Zero-Shot Policy Optimization 3. Learning to Cooperate with Humans using Generative Agents 4. Personalizing Reinforcement Learning from Human Feedback with Variational Preference Learning Find more details on each paper and where to find us in this thread (1/6)

Abhishek Gupta

10,777 views • 1 year ago

Reinforcement learning is used to speed the production of behavior for the Boston Dynamics Atlas humanoid robot. At the heart of the learning process is a physics-based simulator that generates training data for a variety of maneuvers.

Reinforcement learning is used to speed the production of behavior for the Boston Dynamics Atlas humanoid robot. At the heart of the learning process is a physics-based simulator that generates training data for a variety of maneuvers.

RAI Institute

76,553 views • 1 year ago

Wow. This is crazy. A developer trained an AI agent in simulation and deployed it onto a real robotic air hockey table using reinforcement learning. This robot can track the puck with millimeter-level accuracy and react in roughly 20 milliseconds, fast enough to challenge even skilled human players. We’re moving from robots that follow programmed rules to machines that learn strategies in simulation and execute them in the physical world.

Wow. This is crazy. A developer trained an AI agent in simulation and deployed it onto a real robotic air hockey table using reinforcement learning. This robot can track the puck with millimeter-level accuracy and react in roughly 20 milliseconds, fast enough to challenge even skilled human players. We’re moving from robots that follow programmed rules to machines that learn strategies in simulation and execute them in the physical world.

SciTech Era

1,580,234 views • 8 days ago

LAUNCHING SOON! Here is the sneak peek of me using BioLLM to trade Perps with my BRAIN using Phoenix SDK on Solana 1.) Trained BioLLM on Neurons 2.) Simulated the brain using a PATENT PENDING method 3.) Tuned an EEG to work like a computer mouse Are you ready? Lock in.

LAUNCHING SOON! Here is the sneak peek of me using BioLLM to trade Perps with my BRAIN using Phoenix SDK on Solana 1.) Trained BioLLM on Neurons 2.) Simulated the brain using a PATENT PENDING method 3.) Tuned an EEG to work like a computer mouse Are you ready? Lock in.

BioLLM

70,682 views • 20 days ago

Helix is now learning directly from human video data We have already trained on data collected in the real world, including Brookfield residential units To our knowledge, this is the first instance of a humanoid robot learning navigation end-to-end using only human video

Helix is now learning directly from human video data We have already trained on data collected in the real world, including Brookfield residential units To our knowledge, this is the first instance of a humanoid robot learning navigation end-to-end using only human video

Figure

46,041 views • 8 months ago

We want to surpass able-bodied human performance with our technology. Using only his mind, here's precision cursor control from Pager (star of Monkey MindPong) achieving 65% and 88% of the median Neuralinker using a mouse. Join us to breakthrough to 110% and beyond!🧠#techtuesday

We want to surpass able-bodied human performance with our technology. Using only his mind, here's precision cursor control from Pager (star of Monkey MindPong) achieving 65% and 88% of the median Neuralinker using a mouse. Join us to breakthrough to 110% and beyond!🧠#techtuesday

Neuralink

580,094 views • 3 years ago

We're building a surgical robot capable of reaching any brain region. The goal: a generalized neural interface to help solve any condition that originates in the brain.

We're building a surgical robot capable of reaching any brain region. The goal: a generalized neural interface to help solve any condition that originates in the brain.

Neuralink

22,068,826 views • 1 month ago

Cursor testing my full brain simulation's learning capabilities code in comment

Cursor testing my full brain simulation's learning capabilities code in comment

echo.hive

51,039 views • 5 months ago

Excited to share Flow Matching Policy Gradients: expressive RL policies trained from rewards using flow matching. It’s an easy, drop-in replacement for Gaussian PPO on control tasks.

Excited to share Flow Matching Policy Gradients: expressive RL policies trained from rewards using flow matching. It’s an easy, drop-in replacement for Gaussian PPO on control tasks.

David McAllister

150,389 views • 10 months ago

What is neuron degeneration? Skyler, a machine learning engineer at Neuralink, shares how brain-computer interfaces could bypass neural pathways that are broken due to ALS, helping restore the ability to speak to those who have lost it.

What is neuron degeneration? Skyler, a machine learning engineer at Neuralink, shares how brain-computer interfaces could bypass neural pathways that are broken due to ALS, helping restore the ability to speak to those who have lost it.

Neuralink

15,511,728 views • 1 month ago

The future of robotics isn’t being built in factories, it’s being trained in simulation MasterBOT’s proprietary AI model is learning in real-time, powered by reinforcement and continuous feedback loops This is how machines learn to think $BOT

The future of robotics isn’t being built in factories, it’s being trained in simulation MasterBOT’s proprietary AI model is learning in real-time, powered by reinforcement and continuous feedback loops This is how machines learn to think $BOT

MasterBOT

51,108 views • 7 months ago

Watch Spot crouch, jump, climb boxes and leap across gaps, controlled by a neural network trained with reinforcement learning (RL) and multi-expert distillation. Multiple expert policies were trained and distilled together into a single policy that was fine tuned to improve performance over diverse terrains. This work was inspired by ANYmal’s parkour capabilities. The neural network processes depth data from Spot's sensors to construct an understanding of the environment.

Watch Spot crouch, jump, climb boxes and leap across gaps, controlled by a neural network trained with reinforcement learning (RL) and multi-expert distillation. Multiple expert policies were trained and distilled together into a single policy that was fine tuned to improve performance over diverse terrains. This work was inspired by ANYmal’s parkour capabilities. The neural network processes depth data from Spot's sensors to construct an understanding of the environment.

RAI Institute

14,481 views • 1 month ago

Besides reading cool papers, my Twitter account is mostly used for catching up Formula 1 news. Very excited about Lando Norris 's great performance recently. Now we combine Formula Racing to AI research: The following video shows we train a Reinforcement Learning policy to drive a Dallara F317 in Monza and Barcelona Circuits autonomously in simulation. You can think of it as optimizing a Qualifying lap: You want the car to run a single lap, no opponents, with the fastest speed, and shortest lap time. We open-source the code to train RL in a racing simulator with Assetto Corsa, a widely deployed platform for esports. We also train with human demonstrations with multiple expert drivers including a professional e-sports driver. Interestingly AI is not beating the professional yet, but close. See our open-sourced code, simulator, and data here:

Besides reading cool papers, my Twitter account is mostly used for catching up Formula 1 news. Very excited about Lando Norris 's great performance recently. Now we combine Formula Racing to AI research: The following video shows we train a Reinforcement Learning policy to drive a Dallara F317 in Monza and Barcelona Circuits autonomously in simulation. You can think of it as optimizing a Qualifying lap: You want the car to run a single lap, no opponents, with the fastest speed, and shortest lap time. We open-source the code to train RL in a racing simulator with Assetto Corsa, a widely deployed platform for esports. We also train with human demonstrations with multiple expert drivers including a professional e-sports driver. Interestingly AI is not beating the professional yet, but close. See our open-sourced code, simulator, and data here:

Xiaolong Wang

32,356 views • 1 year ago