Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
iterative_inference [2018/05/22 23:44]
admin
iterative_inference [2018/12/21 19:13] (current)
admin
Line 56: Line 56:
 https://​arxiv.org/​abs/​1805.08136v1 Meta-learning with differentiable closed-form solvers https://​arxiv.org/​abs/​1805.08136v1 Meta-learning with differentiable closed-form solvers
  
-In this work we propose to use these fast convergent methods as the main adaptation mechanism for few-shot learning. The main idea is to teach a deep network to use standard machine learning tools, such as logistic regression, as part of its own internal model, enabling it to quickly adapt to novel tasks. This requires back-propagating errors through the solver steps.+In this work we propose to use these fast convergent methods as the main adaptation mechanism for few-shot learning. The main idea is to teach a deep network to use standard machine learning tools, such as logistic regression, as part of its own internal model, enabling it to quickly adapt to novel tasks. This requires back-propagating errors through the solver steps. ​http://​www.robots.ox.ac.uk/​~luca/​r2d2.html
  
 +https://​www.disneyresearch.com/​publication/​iterative-amortized-inference/​ Iterative Amortized Inference
 +
 +https://​github.com/​joelouismarino/​iterative_inference
 +
 +https://​openreview.net/​forum?​id=HygYqs0qKX ​
 +
 +https://​arxiv.org/​abs/​1706.04008 Recurrent Inference Machines for Solving Inverse Problems
 +
 +We establish this framework by abandoning the traditional separation between
 +model and inference. Instead, we propose to learn both components jointly without the need to define
 +their explicit functional form. This paradigm shift enables us to bridge the gap between the fields
 +of deep learning and inverse problems. A crucial and unique quality of RIMs are their ability to
 +generalize across tasks without the need to retrain. We convincingly demonstrate this feature in our
 +experiments as well as state of the art results on image denoising and super-resolution.
 +
 +https://​arxiv.org/​pdf/​1811.02486.pdf Concept Learning with Energy-Based Models
 +
 +https://​openreview.net/​forum?​id=rkxw-hAcFQ Generating Multi-Agent Trajectories using Programmatic Weak Supervision ​
 +
 +e blend deep generative models with programmatic weak supervision to generate coordinated multi-agent trajectories of significantly higher quality than previous baselines.