MIT: Solving brain dynamics gives rise to flexible #machinelearning models https://
news.mit.edu/2022/solving-b
rain-dynamics-gives-rise-flexible-machine-learning-models-1115
… @pierrepinna @sallyeaves @PawlowskiMario @Xbond49 @gvalan @SpirosMargaris @JimMarous @HaroldSinnott @mikeflache @Shi4Tech @TheRudinGroup @ahier #AI #DeepLearning #Fintech #5G
RESEARCH
-
MIT Brain Dynamics Advances Flexible Machine Learning Models
By
–
-
Liquid Networks: Recurrent Neural Nets with Adaptive Learning
By
–
Liquid networks are a class of recurrent neural nets that learn from data to stay adaptable even after training. They have special internal mechanisms that allow them to do so. I gave a talk about them here which might help understand them better: https://
youtu.be/IlliqYiRhMU -
AI Quality and Production Readiness Challenges
By
–
It may get there over time but for now still a bit hit and miss on quality for actual production level usage where robusta checks and testing are needed.
-
Human-in-the-Loop Approach to Machine Learning Systems
By
–
One way of putting the human back in the loop. // @SnorkelAI // #ArtificialIntelligence #MachineLearning #deeplearning
-
Fast Text-Conditional Discrete Denoising Vector-Quantized Latent
By
–
Fast Text-Conditional Discrete Denoising on Vector-Quantized Latent Spaces Rampas et al.: https://
arxiv.org/abs/2211.07292 #ArtificialIntelligence #DeepLearning #MachineLearning -
Teaching Correlation vs Causation in Schools
By
–
Differentiating causation and correlation Schools should teach this. This would avoid a lot of silly debates in our society Here is a great example: the strong correlation between chocolate consumption & rate of Nobel Prize across different countries Crt:NEJM
#data #science -
Galactica Model Repetition Issues Decoding Parameters Analysis
By
–
I’m a bit wondering why we see so much repetitions in the outputs of Galactica people share. If it comes from the model itself or the demo decoding parameters
-
Reliable Knowledge from Parametric Memory in AI Remains an Open Problem
By
–
Yeah — getting reliable, truthful knowledge entirely from parametric memory seems like an open problem. The near-term solutions all involve grounding with queries to external resources.
-
Language Models Continue Sequences from Prompts, Not Maximize Rewards
By
–
they don't maximize rewards, they are given a prompt (a kind of inception) and continue the sequence
-
Extending LLMs to Vision: Incremental Multimodal Integration with Flamingo
By
–
Extending LLMs from text to vision will probably take time but, interestingly, can be made incremental. E.g. Flamingo (
https://
storage.googleapis.com/deepmind-media
/DeepMind.com/Blog/tackling-multiple-tasks-with-a-single-visual-language-model/flamingo.pdf
… (pdf)) processes both modalities simultaneously in one LLM.