Learning Decision Processes in High-Dimensional Space

Deep neural networks often function as 'black boxes.' This paper introduces 'Neural-Probe,' a framework achieving 94% interpretability without sacrificing accuracy.

SC
Dr. Sarah Chen MIT CSAIL
12.4k reads 142 citations
Trending:
Room-Temp Superconductivity Nature Generative Agents in Simulation ArXiv The End of Moore's Law IEEE Spectrum

Recommended Reading

View all trending
Anthropic Research
14.2k

Constitutional AI: Harmlessness from AI Feedback

"As AI systems become more capable, they must align with human values. We propose a method for training a harmless AI assistant through self-improvement without human labels."

Y. Bai et al.
Nature Physics
8.9k

Room-Temperature Superconductivity: A Re-evaluation

L. Monacelli
Science
6.1k

Global Tipping Points in the Earth System

T. Lenton et al.

Curated Syllabi

View archive

Generative AI & LLMs

Curated by Paperless Editors

142 papers

Climate Engineering

Curated by Paperless Editors

89 papers

Quantum Materials

Curated by Paperless Editors

56 papers

Neuroscience

Curated by Paperless Editors

112 papers

Biology & Medicine

Browse subject
Science
5.2k

The complete sequence of a human genome

Nurk et al.
Nucleic Acids Res.
12.5k

AlphaFold Protein Structure Database: Massive scale

DeepMind
NEJM
3.8k

CRISPR-Cas9 gene editing for sickle cell disease

Frangoul et al.
Nature
2.1k

Mapping the developing human immune system

Suo et al.

Latest Arrivals

View archive

Scaling Laws for Neural Language Models

OpenAI / 1.2k reads

"Empirical investigation into scaling laws..."

Direct Air Capture Efficiency

MIT Energy / 854 reads

"Comparative study of solid sorbent technologies..."

Topological Quantum Computing

Microsoft / 2.1k reads

"Progress report on error rates..."

Plastic-eating enzymes

Portsmouth / 3.4k reads

"Directed evolution improves PET depolymerase..."

Dark Matter Detection

LZ Collab / 5.6k reads

"New constraints from the LUX-ZEPLIN experiment..."