PNAS
AI overview
Sign in and the AI will write an overview from our coverage.
Headlines · 4
- SciMuse puts AI-generated research ideas before expert reviewers
A PNAS study introduced SciMuse, a system that combines a knowledge graph spanning 58 million papers with language models to generate personalized interdisciplinary research ideas. Max Planck research group leaders submitted 4,451 valid blind ratings, with proposals averaging 2.40 out of 5 for interest and 24.9% receiving scores of 4 or 5. In a separate comparison, 32.5% of GPT-5.2 proposals received high scores, compared with 29.1% for proposals from senior physics doctoral students and 25.0% for GPT-4 proposals.
虎嗅 AI · 🔥 0 - GPU solver cuLoRADS cracks 18-year-old quantum ordered-search problem
A Shanghai Jiao Tong University team used the GPU-accelerated solver cuLoRADS to directly solve the full semidefinite program for five quantum ordered-search queries, confirming that the maximum searchable ordered-list length is 7,265 and that the scheme fails at 7,266, according to MIT Technology Review China citing a PNAS paper. The program had grown quadratically with list length for 18 years, beyond the reach of CPU solvers at the time. The paper reports that on a single NVIDIA H100, a set of 10-million-order MaxCut semidefinite relaxations was solved in 10 seconds to one minute, versus tens of hours for previously reported CPU methods.
MIT科技评论中文 · 🔥 0 - MIT researchers show randomly wired networks can self-organize learning machinery
In a paper published Oct. 9 in PNAS, researchers at MIT's McGovern Institute for Brain Research and the K. Lisa Yang Integrative Computational Neuroscience (ICoN) Center describe a network of randomly interconnected, neuron-like units that, governed by shared local plasticity rules such as "neurons that fire together wire together," assembled its own learning motifs and learned pattern recognition and image classification. The team reports that its self-organizing network performed at a similar level to the backpropagation-trained networks it compared against. The authors argue that a network's ability to organize its own learning machinery, without a developer designing it, should be a key measure of biological plausibility, while cautioning that the result does not show the brain actually uses this mechanism.
MIT News · AI · 🔥 0 - Why AI should use more Chernoff faces — and data science more manifold fitting
Mathematician Zhigang Yao responds to a Harvard Data Science Review column by founding editor Xiao-Li Meng: AI should make more use of Chernoff faces, reorganizing high-dimensional information into forms people can perceive, compare and question, while data science should do more manifold fitting — treating the underlying geometry itself as the signal to be estimated. The essay traces a line from principal curves and principal flows to principal sub-manifolds, manifold fitting under unbounded noise, and CycleGAN-based manifold fitting, with applications including the single-cell framework scAMF and a study of 251 NMR metabolic biomarkers in 212,853 UK Biobank participants.
虎嗅 AI · 🔥 0
Experience and discussion from the community
Share my PNAS experienceAsk about PNAS
Nobody has shared their experience with PNAS yet.