Principal Investigator. Institute of Neuroscience, CAS. Circuit mechanisms for perception, decision making, top-down modulation, and dendritic computation.

Shanghai
Joined May 2011
Ninglong Xu retweeted
Many people are talking about Marco Rubio's speech at the Munich Security Conference a week ago. It may have been the most explicit defense of the colonial tradition by a senior Western official since Winston Churchill and Adolf Hitler’s imperial diatribes nearly a century ago — supremacist views that claimed millions of lives from the fields of Bengal to the streets of Leningrad. But few are comparing it to the speech that came right afterwards — from Chinese Foreign Minister Wang Yi. The two speeches offered perhaps the clearest expression of the shape of the bifurcating world order that we have heard from world leaders in this century. Where Rubio drew a racial line between the governors and the subjects of the world system, Wang Yi outlined principles for global governance rooted in “sovereign equality, international rule of law, multilateralism, a people-centered approach, and real actions”. Where Rubio attacked global institutions, Wang Yi defended the universality of the UN in terms that spoke directly to the imperatives of the Global South: “On the UN platform, each country, regardless of its size or wealth, has a voice and a sacred vote as well as its due obligations and equal rights.” Without the United Nations, he warned, “the world would revert to the law of the jungle where the strong prey on the weak.” Where Rubio called on the West to be unapologetic and unashamed in imposing its way of life on others, Wang Yi called on major countries to lead by example — recognizing global diversity, “pursuing cooperation instead of conflict or confrontation” and "promoting equality instead of imposing their will onto others.” Where Rubio insisted on throwing off the fetters of international law, Wang Yi emphasized the importance of “observing rules instead of applying double standards.” There is hope amid all the gloom. First, because Rubio's speech was obviously that of a hegemony in decline. The past to which he refers is over and cannot be brought back. And second, because there are still adults in the room — whose words, like Rubio's, are not hollow, but reflect the ongoing construction of a fairer new world order.
3
128
4
355
9,889
Ninglong Xu retweeted
Demis Hassabis’s “Einstein test” for defining AGI: Train a model on all human knowledge but cut it off at 1911, then see if it can independently discover general relativity (as Einstein did by 1915); if yes, it’s AGI.
635
770
293
11,538
2,177,725
My postdoc work "Unsupervised pretraining in biological neural networks" is out now, along with more than 400GB (milliions neurons) neural data. nature.com/articles/s41586-0…. Figshare and Github links can be found in the paper.
4
61
2
311
23,366
Free webinar from Dr. Yulong Li: Spying on Neuromodulation with Novel Genetically Encoded Fluorescent Sensors. Feel free to attend and communicate with us. CET 2:00 PM, March 28th (Thursday) bityl.co/Oz7l
21
2
72
7,899
Clone Robotics is a humanoid robot startup that flies under everyone's radar; they have a uniquely innovative approach to the robot's body. The Poland-based company, founded in Nov 2021, is designing a human-like musculoskeletal body using hydraulic tendon muscles.
47
235
54
1,713
599,274
Ninglong Xu retweeted
Elon Musk
@elonmusk15 Jan 2024
Optimus folds a shirt
23,186
30,876
12,055
248,084
76,038,253
Ninglong Xu retweeted
Replying to @erikbryn
Too often, we think a task is easy because some animal can do it. But the reality is that the task is fiendishly complex and the animal is much smarter than we think. Conversely, we think tasks like playing chess, calculating an integral, or producing grammatically correct text are complex because only some humans can do them after years of training. But it turns out these things aren't that complicated and computers can do them much better than us. This is why the phrase "Artificial General Intelligence" to designate human-level intelligence makes absolutely no sense.
53
133
31
1,324
303,483
Excited to introduce our optimized green and red fluorescent dopamine (DA) GRAB sensors #GRABSensors #dopamine Great collaboration with Mitsuko Uchida‘s group at Harvard University, Yingjie Zhu’s group at SIAT, John Williams’ group and Marina Wolf’s group at OHSU.
3
25
1
155
20,389
Ninglong Xu retweeted
This is an ape ("Kanzi") playing Minecraft! A fascinating experiment on non-human biological neural networks 🙉 I've been teaching AI to play Minecraft for too long. There're so many similar techniques that the ape trainers used: - In-context reinforcement learning: Kanzi gets a fruit or peanut whenever he hits a marked milestone in the game, incentivizing him to follow the in-game guides. - RLHF: Kanzi doesn't understand much language, but he can see the trainers cheering him on, and he occasionally cheers them back! That gives him a strong signal that he's on the right track. - Imitation learning: the trainers show Kanzi just 1 demonstration of how to do a task, and he immediately grasps the concepts. It's much more efficient than using rewards alone. - Curriculum learning: they start with very simple environments to gradually teach Kanzi the controls. At the end, Kanzi is able to navigate complex caves, mazes, and the Nether. It also amazes me how strong the ape's vision system is. Kanzi never saw Minecraft in his life, and for sure his ancestors didn't either. Yet he rapidly adapts to Minecraft's texture and physics, which are dramatically different from the natural world. This level of generalization is far beyond what our most powerful vision models can do today. We are right in the thick of Moravec's paradox again: our best AIs are approaching human level on understanding language, but far behind animals on parsing pixels. From YouTube channel "ChrisDaCow": youtube.com/watch?v=UKpFoYqN… Researchers are from the Ape Initiative, a non-profit org.
110
798
215
3,747
746,138
Ninglong Xu retweeted
Our latest paper on trace fear learning is online. Check it out. Learning-prolonged maintenance of stimulus information in CA1 and subiculum during trace fear conditioning sciencedirect.com/science/ar…
3
7
1,083
The team work about Single-cell spatial transcriptome reveals cell-type organization in the macaque cortex. Please refer to: doi.org/10.1016/j.cell.2023.…
4
2
23
1,737
(1/3) Exciting news! Introducing highly selective next-generation norepinephrine (NE) GRAB sensors! Our optimized fluorescent NE sensors possess 1000-fold selectivity over DA and a 4-fold increase in behavior-stimulated in vivo signals compared to previous prototypes.
2
71
338
53,025
Students don't need a perfect teacher. They need a happy teacher, who's gonna make them excited to come to school and grow a love for learning.
210
4,138
307
22,861
1,642,915
Ninglong Xu retweeted
Cats and dogs understand how the world works and don't have language. LLM may help but are clearly not necessary. Current approaches to model-based RL are insufficient.
31
18
5
309
37,778
Ninglong Xu retweeted
To those who think AGI is upon us: 1. why don't we have level-5 autonomous driving? Any 17 year old can learn to drive in 20 hours of training. 2. Why don't we have domestic robots that can clear the dinner table and fill the dishwasher? Any 10 year old can learn to do that in minutes. Obviously, we are still missing something big for machines to reach human-level intelligence. How do we get machines to learn how the world works?
891
472
132
3,775
1,731,601
Sneak peek of our symposium @IBROorg #IBRO2023 with @MMaravall. We will be discussing 'Cognitive Computations in the Sensory Cortices' with @timbuschman @ninglong @k47h4 and @TouchMoveLab member tweet-less Isabelle Ferezou 🇪🇸 ¡Ole!
1
4
1
14
1,818
Spontaneous (unrehearsed) activity of the remarkable Dendritic Tree-o of @Ryohei_Neuro, @mattlark and @FritjofHelmchen at the 2023 #Dendrites GRC
3
7
1
65
10,470
BTCC Talk Series: Lusha Zhu Principal Investigator at School of Psychological and Cognitive Sciences, IDG/McGovern Brain Institute, Peking University Structure and influence in an interconnected world Apr 5, 2023 10:00- AM (JST) Registration: us02web.zoom.us/webinar/regi…
2
9
1
15
7,129
Komiyama lab statement on An Wu.
26
72
8
506
276,991
Glad that it's now out in Journal of Neuroscience, see our final version here: jneurosci.org/content/42/49/… (no paywall). Am again grateful to @drkjjeffery and @rmgrieves for making this possible. Also thank a couple of anonymous reviewers for their advice.
How does RSC support orientation in complex spaces? Rats navigated in different environments, multidirectional cells encoded 4 local directions in a 4-box, 2 opposite directions in a 2-box and sometimes 1 direction in 1-box, while HD cells always showed 1 global direction. 1/
3
6
32