正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

#SIGGRAPHAsia2024 🔥 Replace your Marching Cubes code with our Occupancy-Based Dual Contouring to reveal the "real" shape from either a signed distance function or an occupancy function. No neural networks involved. Web:

Minhyuk Sung

1,982 subscribers

14,627 次观看 • 1 年前 •via X (Twitter)

教育游戏科学技术 #SIGGRAPHAsia2024

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

RealityKit experiment #75 (gist in the replies) LowLevelMesh shape shifting by interpolating between two signed distance fields and meshing with marching cubes

RealityKit experiment #75 (gist in the replies) LowLevelMesh shape shifting by interpolating between two signed distance fields and meshing with marching cubes

Matt Pfeiffer ᯅ

22,946 次观看 • 9 个月前

I've been digging into what Roblox is working on with Terrain. There's an ongoing effort to switch from an occupancy model to a SDF model (signed distance field) In practice voxels aren't stored as a [0 to 1] value anymore, it's a [-1 to 1] value.

I've been digging into what Roblox is working on with Terrain. There's an ongoing effort to switch from an occupancy model to a SDF model (signed distance field) In practice voxels aren't stored as a [0 to 1] value anymore, it's a [-1 to 1] value.

Max ¯\_(ツ)_/¯

17,822 次观看 • 2 个月前

Andrej Karpathy summarizes how neural networks function, including the process of using a loss function in training and applying backpropagation for optimizing the network parameters through gradient descent.

Andrej Karpathy summarizes how neural networks function, including the process of using a loss function in training and applying backpropagation for optimizing the network parameters through gradient descent.

tetsuo

141,964 次观看 • 7 个月前

How to insert images from the web using the IMAGE function. 📸

How to insert images from the web using the IMAGE function. 📸

Excel Dictionary

77,333 次观看 • 1 年前

🔥 TypeScript Tip 🔥 Here, we build a function that creates zod schemas from your own interfaces, with full autocomplete. The lesson? It's often really hard to get autocomplete when your function _accepts the results of another function_.

🔥 TypeScript Tip 🔥 Here, we build a function that creates zod schemas from your own interfaces, with full autocomplete. The lesson? It's often really hard to get autocomplete when your function _accepts the results of another function_.

Matt Pocock

49,909 次观看 • 3 年前

NOKIA really is that pre-function, during the FUNCTION and post function tune for real. Low key a hype man turned into a song 🔥 🥃 👑

NOKIA really is that pre-function, during the FUNCTION and post function tune for real. Low key a hype man turned into a song 🔥 🥃 👑

Hip-Hop Unison

13,486 次观看 • 1 年前

This man makes an astute observation at the gym Based on the gym time occupancy rates, he believes the rate of unemployment is much higher

This man makes an astute observation at the gym Based on the gym time occupancy rates, he believes the rate of unemployment is much higher

Financial Dystopia

232,725 次观看 • 1 年前

Ryobi dual function seat 📹 Krugerconstruction

Ryobi dual function seat 📹 Krugerconstruction

Science girl

246,627 次观看 • 11 个月前

If gyms had machines that involved games like this, their occupancy rates would probably skyrocket.

If gyms had machines that involved games like this, their occupancy rates would probably skyrocket.

Dudes Posting Their W’s

1,954,425 次观看 • 2 个月前

Individual Neuron: Neural Network A neuron in a neural network performs a weighted sum of inputs, adds a bias, and applies an activation function like sigmoid, ReLU, or tanh, introducing non-linearity. This output helps the neuron learn and represent patterns in the data.

Individual Neuron: Neural Network A neuron in a neural network performs a weighted sum of inputs, adds a bias, and applies an activation function like sigmoid, ReLU, or tanh, introducing non-linearity. This output helps the neuron learn and represent patterns in the data.

ₕₐₘₚₜₒₙ — e/acc

47,467 次观看 • 1 年前

I added a "locking function" that can't evolution of Digimon, and a function to choose whether to return to the egg or maintain the current state when Digimon can no more evolve

I added a "locking function" that can't evolution of Digimon, and a function to choose whether to return to the egg or maintain the current state when Digimon can no more evolve

디지탭 언오프 / DigiTap Unoff

11,015 次观看 • 2 年前

This video, created by my dear coauthor Mahdi E Kahou for our teaching and papers, shows how overparameterized neural networks produce smooth function approximations even in the context of the Runge phenomenon. Some background. Imagine you want to approximate the Runge function using polynomial interpolation at equally spaced points. It is well known that, despite targeting an infinitely differentiable function, such a polynomial approximation produces oscillatory behavior that worsens with the degree of the polynomial. In other words, higher-degree polynomial approximations might not improve accuracy. Instead, approximate the Runge function with a neural network (here, two layers are just to make the example concrete; nothing fundamental depends on it). As you increase the number of parameters well above the 11 training points (in our example, a two-layer neural network with 128 nodes each), you nicely converge to the target, without wild oscillations. Yes, this has much to do with double descent and benign overparameterization, but the main punchline of this post is that neural networks are really very different types of animals than polynomial approximations. And yes, Chebyshev nodes and splines exist, and in this case, they will prevent the oscillations. But that's not the point. Chebyshev nodes and splines still confront Faber’s theorem, which states that for any system of polynomial interpolation nodes, there exists a continuous function whose sequence of interpolating polynomials diverges as the number of nodes grows to infinity. Faber’s theorem does not apply to neural networks because they are not polynomials. The notebook, if you want to check the details, is here: Stay tuned for more on this 👀

This video, created by my dear coauthor Mahdi E Kahou for our teaching and papers, shows how overparameterized neural networks produce smooth function approximations even in the context of the Runge phenomenon. Some background. Imagine you want to approximate the Runge function using polynomial interpolation at equally spaced points. It is well known that, despite targeting an infinitely differentiable function, such a polynomial approximation produces oscillatory behavior that worsens with the degree of the polynomial. In other words, higher-degree polynomial approximations might not improve accuracy. Instead, approximate the Runge function with a neural network (here, two layers are just to make the example concrete; nothing fundamental depends on it). As you increase the number of parameters well above the 11 training points (in our example, a two-layer neural network with 128 nodes each), you nicely converge to the target, without wild oscillations. Yes, this has much to do with double descent and benign overparameterization, but the main punchline of this post is that neural networks are really very different types of animals than polynomial approximations. And yes, Chebyshev nodes and splines exist, and in this case, they will prevent the oscillations. But that's not the point. Chebyshev nodes and splines still confront Faber’s theorem, which states that for any system of polynomial interpolation nodes, there exists a continuous function whose sequence of interpolating polynomials diverges as the number of nodes grows to infinity. Faber’s theorem does not apply to neural networks because they are not polynomials. The notebook, if you want to check the details, is here: Stay tuned for more on this 👀

Jesús Fernández-Villaverde

46,527 次观看 • 27 天前

Neural Software talk I gave at GTC this year. The demo from this presentation is a basic Neural OS I wrote in July of 2024 inspired by Andrej Karpathy's 'LLM OS' post. Lambda's secret mission is to replace all of world's software with neural networks.

Neural Software talk I gave at GTC this year. The demo from this presentation is a basic Neural OS I wrote in July of 2024 inspired by Andrej Karpathy's 'LLM OS' post. Lambda's secret mission is to replace all of world's software with neural networks.

stephen balaban

17,710 次观看 • 1 年前

Trash To Treasure! DIY Multi-Function Stove From An Old Truck Wheel 🌲🔥

Trash To Treasure! DIY Multi-Function Stove From An Old Truck Wheel 🌲🔥

Gisa w'I Rwanda🇷🇼❤️🇷🇼

27,779 次观看 • 12 天前

🚨RFK JR: "Your whole industry told the public for years to 'trust the experts.' 'Trust the experts' is not a function of science or democracy, it's a function of totalitarianism and religion."

🚨RFK JR: "Your whole industry told the public for years to 'trust the experts.' 'Trust the experts' is not a function of science or democracy, it's a function of totalitarianism and religion."

Autism Capital 🧩

180,974 次观看 • 1 年前

Introducing: Function Scans. Starting with the first 22-minute Full Body MRI with FDA-cleared AI. From 60 minutes to 22 and $1,500 to $499. To help detect early signs of hundreds of diseases. Now available through Function.

Introducing: Function Scans. Starting with the first 22-minute Full Body MRI with FDA-cleared AI. From 60 minutes to 22 and $1,500 to $499. To help detect early signs of hundreds of diseases. Now available through Function.

Function Health

97,384 次观看 • 1 年前

What we can or cannot do, what we consider possible or impossible, is rarely a function of our true capability. It is more likely a function of our beliefs about who we are.

What we can or cannot do, what we consider possible or impossible, is rarely a function of our true capability. It is more likely a function of our beliefs about who we are.

Tony Robbins

86,029 次观看 • 1 个月前

EEVEE shader-based screen-space point clouds. Done with displacement shaders – vertices expand in the shader. Feels as fast as pure GLSL. Distance scaling uses a power function for smoother falloff with depth. #b3d #blender

EEVEE shader-based screen-space point clouds. Done with displacement shaders – vertices expand in the shader. Feels as fast as pure GLSL. Distance scaling uses a power function for smoother falloff with depth. #b3d #blender

Damjan Minovski

21,168 次观看 • 1 年前