Home▸Physics & Mechanics▸Transfer Learning Lab: 2D

🔁 Transfer Learning Lab: 2D

2D comparison of a scratch network and a pre-trained, partially frozen network on a shifted target task.

Physics & Mechanics2DModerate60 FPS📱 Mobile-adapted⇄ 2D version

Details

CategoryPhysics & Mechanics

RenderingCanvas 2D

DifficultyModerate

Target FPS60

ControlsSet the domain shift (rotation and translation), the number of labelled target points and noise, choose pre-training steps, how many layers are frozen and whether the head is re-initialised, tune learning rate and speed, click layers of the transfer network to freeze them, and drag, add or erase target points.

transfer learningfine-tuningfrozen layersdomain shiftneural networksartificial intelligence

Related simulations

SPH FluidFluid simulation via SPH: pressure, viscosity, gravity and wall bouncing. Splash impulse…

Spring-Mass Cloth SimulationExplore the physics of cloth movement with this simulation! Adjust the spring stiffness and…

Car PhysicsRealistic driving with suspension, drift, nitro and particle effects. Navigate rough terrain. ⊞ All 3207 in Physics & Mechanics →

⚙ Under the hood

Interactive 2D comparison of a network trained from scratch against a pre-trained network with frozen layers, both adapting to a rotated and shifted target task from only a few labelled points.

transfer learningfine-tuningfrozen layersdomain shiftneural networksartificial intelligence

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install