compress566 @compress566

Beijing Joined December 2022

Tweets

25
Followers

13
Following

310
Likes

134

compress566 @compress566

2 years ago

Idempotent Learned Image Compression with Right-Inverse is accepted by NeurIPS 2023. We identify that invertibility of transform is sufficient but not necessary for idempotence. Instead, it can be relaxed into right-invertibility.

0 0 0 379 0

Anian Ruoss @anianruoss

2 years ago

Your LLM is secretly a compressor and you should treat it like one! arxiv.org/abs/2309.10668 Joint work with @gregdeletang, @duquenne_pa, Elliot Catt Tim, Genewein, Christopher Mattern, @jordigraumo, @liwenliang, Matthew Aitchison, @LaurentOrseau, @mhutter42, and Joel Veness

4 30 125 40K 64

compress566 @compress566

2 years ago

arxiv.org/abs/2207.14524 Learned image compression can be very fast: 145 fps for encoding and 208 fps for decoding on a Tesla T4 GPU for 1080p images. On CPU, the latency of our implementation is comparable with JPEG XL.

0 0 4 186 0

compress566 @compress566

2 years ago

arxiv.org/abs/2308.13287 Efficient Learned Lossless JPEG Recompression can obtain a good throughput of 57 FPS for 1080P images on NVIDIA T4 GPU，

0 0 2 74 0

compress566 @compress566

2 years ago

If you are interested in the cross-platform decoding of neural codec, you may want to read 'He, D., Yang, Z., Chen, Y., Zhang, Q., Qin, H. and Wang, Y., 2022. Post-training quantization for cross-platform learned image compression. arXiv preprint arXiv:2202.07513'

0 0 1 237 0

compress566 @compress566

2 years ago

I think JPEG lossless recompression is important but is ignored currently: Guo, Lina, Xinjie Shi, Dailan He, Yuanyuan Wang, Rui Ma, Hongwei Qin, and Yan Wang. "Practical Learned Lossless JPEG Recompression with Multi-Level Cross-Channel Entropy Model in the DCT Domain." CVPR 2022

Joe Karlsson @JoeKarlsson1

2 years ago

0 1 7 1K 0

Download Image

1 2 7 712 0

compress566 @compress566

2 years ago

What's the coding theory behind Maxine? It is not only the rate-distortion. Our paper (arxiv.org/abs/2308.08154) propose a theoretical framework for conditional perceptual quality compression, which fits well with Maxine (the landmark as Y, the input GOP as X).

Matesanz @AIMatesanz

5 years ago

3 104 133 0 3

Download Video

0 1 2 306 0

Jim Fan @DrJimFan

2 years ago

There're few who can deliver both great AI research and charismatic talks. OpenAI Chief Scientist @ilyasut is one of them. I watched Ilya's lecture at Simons Institute, where he delved into why unsupervised learning works through the lens of compression. Sharing my notes: -…

55 421 3K 820K 3K

Download Video

Stephan Mandt @ AISTATS’25 @StephanMandt

2 years ago

@DrJimFan @ilyasut Given the recent hype of talking about "compression" in mainstream ML, I feel I need to shamelessly advertise our recent survey on neural data compression--a field that has existed for several years: nowpublishers.com/article/Detail…

0 1 14 767 1

compress566 @compress566

2 years ago

Trial-and-error is frustrating😂. Learning more math now

0 0 0 55 0

Stephan Mandt @ AISTATS’25 @StephanMandt

2 years ago

Excited to speak at Columbia University’s LEAP center for climate science today. See Zoom details below. 🌍🤖

LEAP-STC @LeapStc

2 years ago

Excited to speak at Columbia University’s LEAP center for climate science today. See Zoom details below. 🌍🤖

1 4 6 3K 0

Download Image

0 1 17 1K 0

Ravid Shwartz Ziv @ziv_ravid

2 years ago

After a heated debate with @ylecun (don't worry, no researchers were harmed), we need your input! In self-supervised learning, what should you do with the information of the input?🧠 🅰️ Compressing 🅱️ Not compressing Get your votes in & let's settle this once and for all! 🗳️🎉

40 6 40 161K 31

compress566 @compress566

2 years ago

Bit Allocation using Optimization is accepted by ICML 2023, Arxiv:arxiv.org/abs/2209.09422 Code: github.com/tongdaxu/Bit-A…

0 0 2 206 0

Dinghuai Zhang 张鼎怀 @zdhnarsil

3 years ago

Our #ICML2023 workshop proposal "Structured Probabilistic Inference & Generative Modeling" has been accepted 🎉. We can't wait to engage in insightful discussions with experts in probabilistic ML and other areas at the beautiful Hawaii 🌴🏖️. Check🔍: spigmworkshop.github.io

5 24 154 42K 22

AppleInsider @appleinsider

2 years ago

Apple acquires AI-powered video compression startup WaveOne appleinsider.com/articles/23/03… #Apple

0 7 33 13K 1

Stephan Mandt @ AISTATS’25 @StephanMandt

2 years ago

🎉Exciting news! Our "Neural Compression" workshop proposal has been accepted at #ICML 2023! Join us to explore the latest research developments, including perceptual losses and more compute-efficient models! @BerivanISIK, @YiboYang, @_dsevero, @karen_ullrich, @robamler

4 24 96 40K 9

compress566 @compress566

3 years ago

What can compression learn from AlphaZero and ChatGPT? (1) AlphaCodec: Discovering better coding tools/codec with reinforcement learning instead of human efforts (2) RLHF (reinforcement learning from human feedback) for visual quality optimization What do you think?

0 0 1 71 0

Stephan Mandt @ AISTATS’25 @StephanMandt

3 years ago

My talk on the challenges of practical #NeuralCompression is now on Youtube: youtube.com/watch?v=v6DWk7…. I still can't believe it's not better compressed 😉 Thanks, @ICBINBWorkshop, for hosting me!