Introducing Neural Developmental Programs (NDPs)🧬🧠Instead of neural networks with... fixed architectures, we allow neural networks to grow through a dynamic self-organizing process, inspired by how biological nervous systems develop👇 PDF:show more

Visualization of what is inside of AI models. This... represents the layers of interconnected neural networks. And yes patterns do develop and they can form a signature of how they think. The pattern can be seen as the thought process.show more

Imagine: Parallel Neural Networks with Shaders 🤔 (GLSL instead... of CUDA) Parallelization by encoding the weights and activations as Buffer Objects or Textures, using fs to perform matmul and activation functions in parallel. Each fragm shader instance parproc 1..n neuron.show more

NeuRBF: A Neural Fields Representation with Adaptive Radial Basis... Functions paper page: present a novel type of neural fields that uses general radial bases for signal representation. State-of-the-art neural fields typically rely on grid-based representations for storing local neural features and N-dimensional linear kernels for interpolating features at continuous query points. The spatial positions of their neural features are fixed on grid nodes and cannot well adapt to target signals. Our method instead builds upon general radial bases with flexible kernel position and shape, which have higher spatial adaptivity and can more closely fit target signals. To further improve the channel-wise capacity of radial basis functions, we propose to compose them with multi-frequency sinusoid functions. This technique extends a radial basis to multiple Fourier radial bases of different frequency bands without requiring extra parameters, facilitating the representation of details. Moreover, by marrying adaptive radial bases with grid-based ones, our hybrid combination inherits both adaptivity and interpolation smoothness. We carefully designed weighting schemes to let radial bases adapt to different types of signals effectively. Our experiments on 2D image and 3D signed distance field representation demonstrate the higher accuracy and compactness of our method than prior arts. When applied to neural radiance field reconstruction, our method achieves state-of-the-art rendering quality, with small model size and comparable training speed.show more

Inspired by the Tesla Optimus video released yesterday, we... made some videos of what our robots at @1x__tech can do! Thread: 1/4 The behavior you see here is controlled end-to-end from pixels->actions through a single neural net, at 1X speedshow more

🏠 How can AI understand the structure of an... entire building from images? PolyLayout introduces a new approach for multi-room 3D layout estimation — allowing AI to jointly reconstruct connected indoor spaces instead of treating every room independently. By combining neural networks with explicit geometric reasoning, PolyLayout predicts more accurate and robust room layouts while adapting to different scenes and camera setups. A step closer to machines that can truly understand the spaces around us. 🧠🏗️ Paper Title: PolyLayout: Multi-room Manhattan Layout Estimation Project: Link:show more

🚨🇷🇺🇨🇳 Russia and China develop breakthrough AI-powered earthquake forecasting... method The new AI model predicts earthquakes by tracking changes in the Earth’s crust with superior accuracy—at a fraction of traditional computing costs. 🔸 Geoacoustic emissions are used — sounds produced as tectonic stress builds up underground — to pinpoint possible earthquake precursors 🔸The new system combines Physics-Informed Neural Networks (PINNs) with Kolmogorov-Arnold Networks (KANs) to process seismic data 🔸 The team proposes using human-made noise as seismic sonar to reveal underground details that natural signals miss 🔸 The approach slashes computing costs while delivering sharper 2D and 3D geological maps Beyond earthquake forecasting, the same technology could also aid mineral exploration and infrastructure planning—by revealing subsurface structures with unprecedented clarityshow more

Your brain physically rewires itself every time you think... a thought. Donald Hebb stumbled onto this principle in 1949 while studying memory formation in lab rats. He noticed something that should have been impossible: neurons that activated simultaneously began forming stronger connections over time, creating dedicated pathways where none existed before. Scientists called it Hebb's Law. The rest of us call it "neurons that fire together wire together." What Hebb discovered wasn't just a mechanism for learning. He had found the biological foundation of human transformation. Every habit, every skill, every automatic response in your body exists as a neural pathway carved by repetition. The route from your bedroom to your kitchen becomes a superhighway in your brain because you walk it every morning. The sequence of movements you use to tie your shoes becomes hardwired because you've done it thousands of times. But, this same process builds your personality. That tendency to check your phone when you feel anxious? Neural pathway. The automatic urge to argue when someone challenges your opinion? Neural pathway. The way you deflect compliments or seek validation or avoid difficult conversations? All neural pathways, strengthened every time you repeat the pattern. Your brain cannot distinguish between physical actions and mental habits. Both carve grooves in your neural architecture. Both become automatic responses when triggered. Both feel like "who you are" because they happen without conscious choice. But, most people spend decades accidentally building neural superhighways to behaviors they claim they want to change. You say you want to be confident, then practice self doubt every day. You say you want to be productive, then strengthen procrastination pathways by checking social media when work feels hard. You say you want authentic relationships, then wire yourself for people pleasing by avoiding conflict whenever it arises. The brain observes your actions and assumes this must be what you want. So it builds infrastructure to make these patterns easier to execute in the future. Neuroplasticity research reveals something most people find deeply unsettling: there is no "fixed self." The personality you think defines you is just a collection of neural pathways that have been reinforced more often than others. The pathways you travel most frequently become the widest roads. The thoughts you think most often become the loudest voices. The behaviors you repeat most consistently become your automatic responses. But the same mechanism that locks you into patterns can unlock you from them. Every time you catch yourself mid pattern and choose differently, you send a signal to your brain that the old pathway might not be serving you anymore. Every time you practice a new response instead of defaulting to the familiar one, you begin building new neural infrastructure. The process feels awkward at first because you're literally walking through mental wilderness, creating trails where no trails existed. But repetition turns trails into paths, paths into roads, roads into superhighways. This is why changing habits through willpower alone fails. You're trying to muscle through established neural superhighways instead of building alternative routes. The old pathways don't disappear just because you want them to. They have to be replaced through deliberate rewiring. The most sophisticated meditation practitioners in the world understand this intuitively. They don't just sit quietly hoping for peace. They systematically rewire their brains by repeatedly choosing calm responses instead of reactive ones. Ten thousand hours of practice creates neural pathways so robust that serenity becomes their default state. Professional athletes do the same thing with performance. They don't just practice their sport. They practice the mental patterns that support excellence until confidence, focus, and resilience become neurologically hardwired. The implications of neuroplasticity extend far beyond personal development. Every social bias, every cultural assumption, every automatic judgment you make exists as neural wiring built through repetition. The way you unconsciously categorize people, the assumptions you make about different groups, the stereotypes that feel "obviously true" are all learned pathways that can be unlearned. Societies change when enough individuals rewire their neural patterns around new ways of thinking and behaving. The brain you have right now is not the brain you're stuck with. It's the brain you've trained through repetition. Every thought you choose, every action you take, every response you practice is a vote for the kind of neural architecture you want to build. Most people cast these votes unconsciously, then wonder why their life feels automatic and unchangeable. The moment you realize you're the architect of your own neural patterns is the moment real transformation becomes possible. Your neurons are firing right now as you read this. What are you choosing to wire them toward?show more

Introducing Kaleido💮 from AI at Meta — a universal... generative neural rendering engine for photorealistic, unified object and scene view synthesis. Kaleido is built on a simple but powerful design philosophy: 3D perception is a form of visual common sense. Following this idea, we formulate rendering purely as a sequence-to-sequence generation problem, successfully unifying neural rendering with the architecture principles behind modern language and video models. Unlike traditional neural rendering methods, Kaleido learns 3D purely in a data-driven way, without explicit 3D representations or structures. It acquires spatial understanding directly through large-scale video pretraining, then multi-view 3D data finetuning, inspired by how LLMs acquire textual common sense from large corpora before specialising in domains like coding. Through extensive ablations, we progressively modernised the architecture design and training strategies and tackled key scaling challenges in sequence-to-sequence generative rendering, arriving at a design that’s simple, versatile, and scalable. Kaleido significantly outperforms prior generative models in few-view settings, and remarkably is the first zero-shot generative method matches InstantNGP-level rendering quality in multi-view settings. We view Kaleido also as an alternative step towards world modeling that flexibly spans a spectrum of “realities": with many views, it faithfully reconstructs grounded reality; with fewer views, it imagines plausible unseen details. 🔗 Explore more results and paper:show more

Ever wondered if you could extract capabilities and behaviors... from neural networks and reuse/update/route it as needed? We introduce low-rank circuit conditioning, a novel approach that preserves the model's output behavior while reshaping how an existing capability is represented. In the base model, standard compact recovery stalls at 29%. After conditioning, the same extraction pipeline reaches 91.33% autoregressive full-answer recovery from 5.05% of MLP channels. The evidence points to a possibility of extracting and using isolated capabilities saving cost, latency and high adaptability. Read our work to understand more -show more

How does sensorimotor (S1/M1) cortex support adaptive motor control?... Come find out in our latest preprint, which spans the development of a full adult forelimb model + physics simulations, neural-modeling for control, complex 🐭behavior 🕹️, large-scale imaging, and of course DeepLabCut 🦄 and Bernhard Rackl! We hypothesized that S1 supports motor learning by computing prediction errors. To tackle this, we needed to understand what is being represented, and no studies have reported what forelimb S1 represents during learning in mice🧠🐭. Moreover, this requires modeling the body🦾: kinematics, torques, force, muscle activations, & proprioception (muscle spindles & GTOs). After our 7 year journey, we have an answer: S1 & M1 represent muscle-level features. During learning, computational motifs map to functional types (like muscle-encoding), and neural dynamics in S1 change & encode sensorimotor prediction errors! 🧵👇show more

🚨 Breaking: 🇰🇵 North Korea: “Any attempt to intercept... Iranian vessels will be met with direct military protection from North Korean submarines. If anyone tries to seize one of these ships, we will respond by targeting your entire naval fleet.” 🇮🇱 Israel: “Any vessel carrying illegal weapons to terrorist networks will be considered a legitimate target and destroyed. We will not allow a growing Pyongyang–Tehran partnership to develop into a threat to Israel’s security.”show more

‼️ Ukrainian animators rushed into hand-to-hand combat to retake... Krasnoarmeysk (Pokrovsk), which Ukraine has lost forever. Using neural networks, they generated a video claiming that the Russians allegedly altered the Ukrainian Armed Forces' video from Krasnoarmeysk by "pasting" a Russian flag instead of the Ukrainian one. ❗️ In reality, it is quite the opposite. Note the lively movement of cars with headlights on in the Ukrainian version of the video. This is now in a city where a drone can come from anywhere! Yes, and overall the original Russian video is much longer than the fragment that the makeshift AI could "digest." Strangely, the Ukrainian "victors" did not think to run a trolleybus in Pokrovsk – the townspeople have long dreamed of it. - UARUshow more

For most of neuroscience history (until the late 1980s/early... 1990s), neurons were treated as the brain’s only decision-makers. That idea is changing. Astrocytes, once considered passive support cells, are now recognized as global regulators of brain state. They don’t encode individual thoughts or actions. Instead, they integrate activity across huge numbers of synapses and adjust how neural networks behave over time. By releasing modulatory signals and tracking slow changes in activity, astrocytes influence: -Alertness vs. fatigue -Stress vs. calm -Motivation vs. disengagement Importantly, they can shift brain function without rewiring neurons, by tuning the environment neurons operate in. The brain isn’t just wired. It’s regulated, and astrocytes play a central role in that regulation.show more

TESLA’S OPTIMUS BOT: COOKS, CLEANS, AND WILL CHANGE THE... FUTURE FOR THE BETTER Tesla’s humanoid robot, Optimus, isn’t just walking—it’s working. Designed to handle chores like cooking, vacuuming, and folding laundry, Optimus is Tesla’s leap into domestic automation. With sensors, neural networks, and a frame built for agility, it can navigate cluttered homes and messy kitchens while improving its skills over time. More than a sci-fi showpiece, it’s being trained to adapt to household routines. Think of it as a second set of hands—unfazed by spills or socks. It’s here, and it’s learning to make dinner. Source: Nic Cruz Patane Tesla Optimusshow more

You’re looking at neurons growing and connecting in real... time 🧠. A 65-hour recording of hippocampal activity in a rat brain. The hippocampus plays a crucial role in memory and learning. In this footage, neurons extend their dendrites and axons, building and reshaping connections across days. Capturing this process live offers a rare view into how neural circuits form and reorganize. Why this matters ⬇️ 1️⃣It’s a continuous, multi-day recording of living hippocampal neurons under the microscope 2️⃣You can clearly see dendritic branching and network formation 3️⃣It reveals the dynamic processes that drive brain development and plasticity Observing growth at this resolution helps researchers understand how neurons connect—and how disruptions in these processes might contribute to neurological or psychiatric conditions. Credit to Louis Romet and Dr. Christophe Leterrier for the videoshow more

Bio-inspired #TrueAI continues the journey, today we Jose Sánchez... & David Vivancos - e/acc are very glad to introduce for Qubic #OpenScience #MultiNeuraxon 2.0 hibridized with #Aigarth Come-from-Beyond Code as allways at GitHub Demo and #BrainBuilder at Hugging Face Demo Video Explainer later today. Paper will be presented in the following months, stay tuned for updates. 🧠💻Why does it matter? It bridges the gap between artificial and biological intelligence by replacing rigid, layer-by-layer AI pipelines with interconnected neural modules (spheres) that function like distinct regions of the brain. By utilizing continuous-time processing and trinary logic (excitatory, neutral, and inhibitory states), it paves the way for energy-efficient AI capable of real-time adaptation and lifelong learning without suffering from catastrophic forgetting. Stay tuned evolution towards #AGI just started...show more

3D Gaussian Splatting for Real-Time Radiance Field Rendering paper... page: Radiance Field methods have recently revolutionized novel-view synthesis of scenes captured with multiple photos or videos. However, achieving high visual quality still requires neural networks that are costly to train and render, while recent faster methods inevitably trade off speed for quality. For unbounded and complete scenes (rather than isolated objects) and 1080p resolution rendering, no current method can achieve real-time display rates. We introduce three key elements that allow us to achieve state-of-the-art visual quality while maintaining competitive training times and importantly allow high-quality real-time (>= 30 fps) novel-view synthesis at 1080p resolution. First, starting from sparse points produced during camera calibration, we represent the scene with 3D Gaussians that preserve desirable properties of continuous volumetric radiance fields for scene optimization while avoiding unnecessary computation in empty space; Second, we perform interleaved optimization/density control of the 3D Gaussians, notably optimizing anisotropic covariance to achieve an accurate representation of the scene; Third, we develop a fast visibility-aware rendering algorithm that supports anisotropic splatting and both accelerates training and allows realtime rendering. We demonstrate state-of-the-art visual quality and real-time rendering on several established datasets.show more