Gary MarcusThe world needs an international agency for artificial intelligence.
Now. (Or better yet: six months ago)
By Invitation essay in Economist
| https://www.linkedin.com/in/alexander-granderath-35a5a4/ |
Alexander Granderath ✅
- 31 Followers
- 123 Following
- 136 Posts
Supervisory Board Member, Tech Evangelist, Business Angel, Consulting, Investing, Learning together, always good for a smile.
Fahim Farook"Consistency Models. (arXiv:2303.01469v1 [cs.LG])" — A new family of generative models that achieve high sample quality without adversarial training that supports fast one-step generation by design.
Paper: http://arxiv.org/abs/2303.01469
#AI #CV #NewPaper #DeepLearning #MachineLearning
<<Find this useful? Please boost so that others can benefit too 🙂>>
Samples generated by EDM (top),…
Paper: http://arxiv.org/abs/2303.01469
#AI #CV #NewPaper #DeepLearning #MachineLearning
<<Find this useful? Please boost so that others can benefit too 🙂>>
Samples generated by EDM (top),…
Consistency Models
Diffusion models have significantly advanced the fields of image, audio, and video generation, but they depend on an iterative sampling process that causes slow generation. To overcome this limitation, we propose consistency models, a new family of models that generate high quality samples by directly mapping noise to data. They support fast one-step generation by design, while still allowing multistep sampling to trade compute for sample quality. They also support zero-shot data editing, such as image inpainting, colorization, and super-resolution, without requiring explicit training on these tasks. Consistency models can be trained either by distilling pre-trained diffusion models, or as standalone generative models altogether. Through extensive experiments, we demonstrate that they outperform existing distillation techniques for diffusion models in one- and few-step sampling, achieving the new state-of-the-art FID of 3.55 on CIFAR-10 and 6.20 on ImageNet 64x64 for one-step generation. When trained in isolation, consistency models become a new family of generative models that can outperform existing one-step, non-adversarial generative models on standard benchmarks such as CIFAR-10, ImageNet 64x64 and LSUN 256x256.
taz
Hassan SalemHi All, I’m excited to finally join the conversation at #ScienceMastodon.
Here is a belated #introduction. I’m an #evolutionary #biologist currently leading a research group at the Max Planck Institute for Biology. Our lab studies symbioses and the ways by which they facilitate #herbivory in insects, using tortoise beetles as a model.
Looking forward to connecting with many of you here! #evolution #ecology #microbiology #entomology #microscopy
FediTips has moved!If you're @-ing someone on here, it's really important to make sure you've chosen the right person.
Account addresses on Mastodon and the Fediverse always have two parts, the username and the server. Both these parts have to be correct for it to reach the right account.
This is similar to email addresses: you can't just put the first part, you need to include the server at the end too.
To see someone's full account address, look at their profile page. More info at https://fedi.tips/how-to-use-mastodon-and-the-fediverse-basic-tips/#WhatsMyAccountsAddress
Carl T. Bergstrom
Mark RiedlI think the Tesla website is telling me that Cybertrucks should be thrown into black holes
Fahim Farook"StitchNet: Composing Neural Networks from Pre-Trained Fragments. (arXiv:2301.01947v1 [cs.LG])" — A neural network creation paradigm that stitches together fragments (one or more consecutive network layers) from multiple pre-trained neural networks without any retraining.
Paper: http://arxiv.org/abs/2301.01947
#AI #CV #NewPaper #DeepLearning #MachineLearning
<<Find this useful? Please boost so that others can benefit too 🙂>>
StitchNet: Composing Neural Networks from Pre-Trained Fragments
We propose StitchNet, a novel neural network creation paradigm that stitches together fragments (one or more consecutive network layers) from multiple pre-trained neural networks. StitchNet allows the creation of high-performing neural networks without the large compute and data requirements needed under traditional model creation processes via backpropagation training. We leverage Centered Kernel Alignment (CKA) as a compatibility measure to efficiently guide the selection of these fragments in composing a network for a given task tailored to specific accuracy needs and computing resource constraints. We then show that these fragments can be stitched together to create neural networks with accuracy comparable to that of traditionally trained networks at a fraction of computing resource and data requirements. Finally, we explore a novel on-the-fly personalized model creation and inference application enabled by this new paradigm. The code is available at https://github.com/steerapi/stitchnet.
heise onlineAutonomes Fahren: ZF zeigt hochautomatisiertes Elektro-Shuttle für Mischverkehr
Der Autozulieferer ZF hat auf der CES die nächste Generation seiner selbstfahrenden Shuttles vorgestellt, die im gemischten Verkehr ohne Fahrer auskommen soll.
#BEEP #Bus #Elektromobilität #Personentransport #Shuttle #Transport #Verkehr #ZF #autonomesFahren #News
