No article text was recorded for this version.
expressivity-learnability gap
Version 2 of 2651 papers811 conceptsfrozen 2026-10-04
Papers in this version
- Blog: Are Efficient Deep Representations Learnable?
- Blog: Synthetic Data Reveals Generalization Gaps in Correlated Multiple Instance Learning
- Blog: Provable Failure of Language Models in Learning Majority Boolean Logic via Gradient Descent
- Blog: Overcoming a Theoretical Limitation of Self-Attention
- Blog: Faithful Relational Reasoning with Region-based Embeddings: Expressivity of Convex Coordinate-wise Models
- Blog: The Expressive Limits of Diagonal SSMs for State-Tracking
- Blog: Can Transformers Learn n-gram Language Models?
- Blog: Is Deeper Better only when Shallow is Good?
- Blog: The Role of Logic and Automata in Understanding Transformers
- Blog: Weisfeiler and Leman go Machine Learning: The Story so far
- Blog: Expressiveness of Rectifier Networks
- Blog: On the Paradox of Learning to Reason from Data
- Blog: Approximation properties of DBNs with binary hidden units and real-valued visible units
- Blog: SGD Learns the Conjugate Kernel Class of the Network
- Blog: The limitations of optimization from samples
- Blog: Probabilistic Deterministic Finite Automata and Recurrent Networks, Revisited
- Blog: Proof of the Theory-to-Practice Gap in Deep Learning via Sampling Complexity bounds for Neural Network Approximation Spaces
- Blog: Expressive power of recurrent neural networks
- Blog: On the hardness of learning under symmetries
- Blog: Is Deeper Better only when Shallow is Good?
- Blog: Expressiveness of Rectifier Networks
- Blog: Deep neural network approximation theory for high-dimensional functions
- Blog: Overcoming a Theoretical Limitation of Self-Attention
- Blog: Expressive power of recurrent neural networks
- Blog: Hardness of Learning Fixed Parities with Neural Networks
- Blog: Lower Bounds on the Expressivity of Recurrent Neural Language Models
- Blog: Expressive Power and Approximation Errors of Restricted Boltzmann Machines
- Blog: The Limitations of Optimization from Samples
- Blog: Convolutional Rectifier Networks as Generalized Tensor Decompositions
- Blog: Universal Approximation Property of Neural Ordinary Differential Equations
- Blog: Constant Bit-size Transformers Are Turing Complete
- Blog: Can Shallow Neural Networks Beat the Curse of Dimensionality? A mean field training perspective
- Blog: What Can Be Learnt With Wide Convolutional Neural Networks?
- Blog: Understanding the Parameter Space Geometry of Transformers Encoding Boolean Functions
- Blog: The gap between theory and practice in function approximation with deep neural networks
- Blog: On the Paradox of Learning to Reason from Data
- Blog: Universal Approximation of Input-Output Maps by Temporal Convolutional Nets
- Blog: On Multi-Layered Connectionist Models: Adding Layers vs. Increasing Width
- Blog: On the Computational Power of Transformers and Its Implications in Sequence Modeling
- Blog: An overview on deep learning-based approximation methods for partial differential equations
- Blog: Efficient Turing Machine Simulation with Transformers
- Blog: Theory-to-Practice Gap for Neural Networks and Neural Operators
- Blog: Why Deep Neural Networks?
- Blog: On the Representational Capacity of Neural Language Models with Chain-of-Thought Reasoning
- Blog: Lower Bounds on the Expressivity of Recurrent Neural Language Models
- Blog: Cognitive Deficit of Deep Learning in Numerosity
- Blog: Memorization in Attention-only Transformers
- Blog: Transformers are Inherently Succinct
- Blog: Attention Mechanism, Max-Affine Partition, and Universal Approximation
- Blog: The Depth-to-Width Interplay in Self-Attention
- Blog: Neural Networks and the Chomsky Hierarchy
- Blog: Understanding the Expressive Power and Mechanisms of Transformer for Sequence Modeling
- Blog: Formal Language Theory Meets Modern NLP
- Blog: Topological Expressivity of ReLU Neural Networks
- Blog: On the Reasoning Abilities of Masked Diffusion Language Models
- Blog: Learning Deep ReLU Networks Is Fixed-Parameter Tractable
- Blog: On the Expressive Power of Self-Attention Matrices
- Blog: $\mathbb{R}^{2k}$ is Theoretically Large Enough for Embedding-based Top-$k$ Retrieval
- Blog: How Far Can Transformers Reason? The Locality Barrier and Inductive Scratchpad
- Blog: Mind The Gap: Deep Learning Doesn't Learn Deeply
Showing 60 of 651 papers.
History & citation
Current · Permanent link
- Logarithmic Depth Scaling for Efficient Function Approximation merged into this direction (54% overlap)
- Universal Approximation and the Learnability Gap in Deep Learning merged into this direction (90% overlap)
Cite this direction
Latest version (a living page):
@misc{lacuna_expressivity_learnability_gap_f9e85ecc,
title = {expressivity-learnability gap},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/expressivity-learnability-gap/rdr_f9e85ecc681d4a2c98ad9a64323ad99c},
note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-22},
}This exact version (version 2, frozen text):
@misc{lacuna_expressivity_learnability_gap_f9e85ecc_v2,
title = {expressivity-learnability gap},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/expressivity-learnability-gap/rdr_f9e85ecc681d4a2c98ad9a64323ad99c/v/2},
note = {Version 2, frozen 2026-10-04. Accessed 2026-11-22},
}