arXiv research

ART improves transfer learning performance with robust theory and methods.

We develop differential algebraic K-theory for rings of integers in number fields and we construct a cycle map from geometrized bundles of modules over such a ring to the differential algebraic K-theory. We also treat some of the foundational aspects of differential cohomology, including differential function spectra a…

We analyze a monetary system of random money transfer on the basis of double entry bookkeeping. Without boundary conditions, we do not reach a price equilibrium and violate text-book formulas of economists quantity theory (MV=PQ). To match the resulting quantity of money with the model assumption of a constant price, w…

Learning transferable knowledge across similar but different settings is a fundamental component of generalized intelligence. In this paper, we approach the transfer learning challenge from a causal theory perspective. Our agent is endowed with two basic yet general theories for transfer learning: (i) a task shares a c…

Analyzes string topology operations using Chen's integrals and homotopy transfer.

Paper constructs a transfer map for codimension 2 submanifolds in higher index theory.

Paper analyzes AIRL in high-dimensional spaces using random matrix theory.

Much attention has been devoted recently to the generalization puzzle in deep learning: large, deep networks can generalize well, but existing theories bounding generalization error are exceedingly loose, and thus cannot explain this striking performance. Furthermore, a major hope is that knowledge may transfer across …

New theory explains when pre-trained models can improve downstream tasks.

New research on limits of transfer learning, proving key selection and dependence requirements.

We propose a general notion of algebraic gauge theory obtained via extracting the main properties of classical gauge theory. Building on a recent work on transferring curved $A_{\infty}$-structures we show that, under certain technical conditions, algebraic gauge theories can be transferred along chain contractions. Sp…

Researchers prove an $L$-theoretic signature transfer in codimension 2.

A new framework for robust transfer learning that avoids negative transfer in domains with unequal information.

We analyze Gibbs-based transfer learning algorithms using information theory.

Residual networks with depthwise hyperparameter scaling transfer optimal hyperparameters across width and depth.

We extend graph neural networks to transfer performance across different input sizes.

Transfer learning has been proven effective when within-target labeled data is scarce. A lot of works have developed successful algorithms and empirically observed positive transfer effect that improves target generalization error using source knowledge. However, theoretical analysis of transfer learning is more challe…

The study analyzes transfer learning using information theory.

Adversarial training enhances model transferability without sacrificing accuracy.

This paper introduces localized discrepancy theories for unsupervised domain adaptation.

This paper extends transfer operator theory to McKean-Vlasov equations.

Study improves model fit by transferring info from related datasets.

CONCERT improves transfer learning by borrowing partial information from auxiliary datasets.

Enhances generative model accuracy through knowledge transfer.

The study analyzes transfer learning in infinite-width neural networks, improving generalization on target tasks.

Transfer learning is a very important tool in deep learning as it allows propagating information from one "source dataset" to another "target dataset", especially in the case of a small number of training examples in the latter. Yet, discrepancies between the underlying distributions of the source and target data are c…

New method upscales models and transfers hyperparameters efficiently.

This work extends GNNs to handle multiple graphs with non-commuting operators, proving transferability.

Novel AMP framework for multi-environment transfer learning.

The paper examines Adaptive Lasso and Transfer Lasso, highlighting their differences and proposing a new method.

A Systemic Optimal Risk Transfer Equilibrium (SORTE) was introduced in: "Systemic optimal risk transfer equilibrium", Mathematics and Financial Economics (2021), for the analysis of the equilibrium among financial institutions or in insurance-reinsurance markets. A SORTE conjugates the classical Bühlmann's notion of a …

Transfer learning is a popular practice in deep neural networks, but fine-tuning of large number of parameters is a hard task due to the complex wiring of neurons between splitting layers and imbalance distributions of data in pretrained and transferred domains. The reconstruction of the original wiring for the target …

Continuity of earthquake flow map transfers Teichmüller dynamics results.

A transfer learning method builds high-dimensional models using disparate datasets.

New methods avoid spectral pollution in transfer operators for accurate analysis.

Proposes improved classification via transfer learning with regularized linear discriminant analysis.

A simple strategy prevents negative transfer in transfer learning.

We prove the Farrell-Jones Conjecture for algebraic K-theory of spaces for virtually poly-Z-groups. For this, we transfer the 'Farrell-Hsiang method' from the linear case to categories of equivariant, controlled retractive spaces.

This paper examines allocation mechanisms in markets with transfer costs, showing how these costs affect economic efficiency.

We construct Hodge filtered cohomology groups for complex manifolds that combine the topological information of generalized cohomology theories with geometric data of Hodge filtered holomorphic forms. This theory provides a natural generalization of Deligne cohomology. For smooth complex algebraic varieties, we show th…

The paper analyzes how combining samples from two tasks can improve performance, especially in high dimensions.

New method uses correlation-ratio for transfer learning, improving target model inference.

This paper analyzes HTL using stability theory for binary classification.

A method for disentangling text representations without supervision.

Paper improves AIRL by enhancing policy imitation and addressing reward recovery issues.

This paper quantifies hyperparameter transfer and finds embedding layer learning rate is key.

The paper proves a generalized inverse function theorem for curved $L_\infty$ spaces.

Understanding the relationship between the structure of light-harvesting systems and their excitation energy transfer properties is of fundamental importance in many applications including the development of next generation photovoltaics. Natural light harvesting in photosynthesis shows remarkable excitation energy tra…