The paper proposes a learning-theoretic perspective on representation alignment.
problem Understanding how AI models' representations become aligned as they scale.
method Reviewing and connecting different notions of alignment, focusing on stitching.
result Relating properties of stitching to kernel alignment of representations.
MKA incorporates manifold geometry into kernel alignment for more robust representation comparison.
problem Inadequate accounting for manifold geometry in kernel alignment metrics.
method Derives a theoretical framework for Manifold Approximated Kernel Alignment (MKA).
result MKA provides a more robust foundation for measuring representations.
Develops an ordinal-similarity framework for scalable and interpretable representation alignment.
problem Measuring representation similarity in large datasets.
method Triplet and Quadruplet Similarity Indices.
result Demonstrates inherent interpretability, robustness to outliers, and computational efficiency.
AI models aligned with human vision perform well on few data tasks.
problem Few-shot learning performance with limited data.
method Information-theoretic analysis and empirical testing of 491 models.
result Highly aligned models show better robustness to attacks and domain shifts.
Alignment of neural network representations is influenced by SNR and sample size.
problem Understanding how neural network representations align across different conditions.
method Controlled training of neural networks on perturbed datasets, analyzing alignment and generalization.
result Alignment varies monotonically with SNR but non-monotonically with sample size, with minimal alignment near the interpolation threshold.
This work optimizes alignment and uniformity of features on a hypersphere for better downstream performance.
problem Improving the performance of contrastive representation learning.
method Identifying and optimizing alignment and uniformity of features on a hypersphere.
result Directly optimizing alignment and uniformity leads to comparable or better performance than contrastive learning.
Recent work on learning multilingual word representations usually relies on the use of word-level alignements (e.g. infered with the help of GIZA++) between translated sentences, in order to align the word embeddings in different languages. In this workshop paper, we investigate an autoencoder model for learning multil…
CKA with Gaussian RBF kernels converges linearly as bandwidth increases.
problem Understanding the behavior of CKA with large bandwidth Gaussian kernels.
method Analyzing the convergence of CKA based on Gaussian RBF kernels in the large-bandwidth limit.
result CKA based on Gaussian RBF kernels converges linearly as bandwidth increases.
Geometric framework detects concept frustration between human concepts and machine representations.
problem Aligning human concepts with machine learning representations.
method Geometric framework and similarity measures for detecting concept frustration.
result Concept frustration affects machine learning model performance and reorganizes learned concept representations.
New estimator corrects bias in CKA for sparsely sampled neurons.
problem Bias in CKA for sparsely sampled neurons.
method Novel estimator that corrects for input and feature sampling.
result Reliable model-to-brain alignment with sparsely sampled neurons.
We examine the influence of input data representations on learning complexity. For learning, we posit that each model implicitly uses a candidate model distribution for unexplained variations in the data, its noise model. If the model distribution is not well aligned to the true distribution, then even relevant variati…
Network alignment is a critical task to a wide variety of fields. Many existing works leverage on representation learning to accomplish this task without eliminating domain representation bias induced by domain-dependent features, which yield inferior alignment performance. This paper proposes a unified deep architectu…
A novel OT-based method for aligning hyperbolic representations.
problem Aligning different hyperbolic representations of hierarchical data.
method Optimal transport (OT) on the Poincaré model of hyperbolic spaces, using gyrobarycenter mapping.
result Both Euclidean and hyperbolic OT-based methods perform similarly in retrieval tasks.
Representational similarity analysis (RSA) has been shown to be an effective framework to characterize brain-activity profiles and deep neural network activations as representational geometry by computing the pairwise distances of the response patterns as a representational dissimilarity matrix (RDM). However, how to p…
Early alignment in neural networks leads to sparse representations but hinders convergence.
problem The implicit bias of gradient descent during early training phases.
method Quantitative description of early alignment phase in small initialisation, one hidden layer networks.
result Early alignment induces a sparse representation but also hinders convergence to global minima.
Proposes COALA method for learning audio representations aligned with tags.
problem Lack of annotated data for high-performance audio representation learning.
method Aligns latent representations of audio and tags using a contrastive loss.
result Audio embedding model captures both acoustic and semantic characteristics.
High-dimensional models pose both safety benefits and risks.
problem Emergent problems in safety alignment due to high-dimensional representations.
method Detailed visualizations and lower-dimensional subspace projections.
result Dimensional reduction preserves safety alignment while avoiding linear jailbreaking.
Proposes a new method for manifold alignment using geometry-regularized twin autoencoders.
problem Traditional MA methods lack out-of-sample extension and real-world applicability.
method Guided representation learning with geometry-regularized twin autoencoders.
result Improves cross-domain generalization and robustness while maintaining alignment fidelity.
The application of machine learning to bioinformatics problems is well established. Less well understood is the application of bioinformatics techniques to machine learning and, in particular, the representation of non-biological data as biosequences. The aim of this paper is to explore the effects of giving amino acid…
LLM trading agents show risk feedback can improve alignment without fine-tuning.
problem Aligning LLM trading agents with financial risk.
method TradeArena testbed, risk reports, execution simulation, memory replay.
result Risk feedback can improve alignment without fine-tuning, but not universally.
Cross-language learning allows us to use training data from one language to build models for a different language. Many approaches to bilingual learning require that we have word-level alignment of sentences from parallel corpora. In this work we explore the use of autoencoder-based methods for cross-language learning …
New framework assesses neural sensitivity to small perturbations.
problem Comparing neural representations' sensitivity to small changes.
method Local decodable information, Fisher information, and projected pullback/Fisher metric.
result Reveals differences in neural sensitivity not captured by activation alignment.
Paper presents a method to align unpaired samples across different modalities.
problem Challenges in collecting paired samples for multimodal representation learning.
method Uses propensity score alignment based on Rubin's framework to estimate a common space for unpaired samples.
result Optimal transport matching significantly improves alignment in real-world data.
We introduce BilBOWA (Bilingual Bag-of-Words without Alignments), a simple and computationally-efficient model for learning bilingual distributed representations of words which can scale to large monolingual datasets and does not require word-aligned parallel training data. Instead it trains directly on monolingual dat…
Partial soft-matching distance improves neural representation comparison by allowing some neurons to remain unmatched.
problem Neural representations are noisy and contain outliers, making traditional matching methods unreliable.
method Extends soft-matching distance to a partial optimal transport setting, allowing some neurons to remain unmatched.
result Partial soft-matching provides robust correspondences that are more reliable under noise and outliers.
iREPA shows spatial structure, not global semantic, drives generation performance in REPA.
problem Understanding what aspect of the target representation matters for generation.
method Empirical analysis of 27 vision encoders, two modifications to REPA.
result Spatial structure, not global semantic, drives generation performance.
Can neural networks learn to compare graphs without feature engineering? In this paper, we show that it is possible to learn representations for graph similarity with neither domain knowledge nor supervision (i.e.\ feature engineering or labeled graphs). We propose Deep Divergence Graph Kernels, an unsupervised method …
COCA refactors training data to identify and erase unsafe concepts in LLMs.
problem Identifying and erasing unsafe concepts in Large Language Models (LLMs) for safety alignment.
method Concept Concentration (COCA) refactors training data with an explicit reasoning process to identify and erase unsafe concepts.
result COCA significantly reduces both in-distribution and out-of-distribution jailbreak success rates while maintaining strong performance on regular tasks.
New framework aligns latent representations over-the-air using intelligent metasurfaces.
problem Heterogeneous transmitter-receiver models produce misaligned latent representations in semantic communication.
method Intelligent metasurfaces (SIM) emulate supervised and zero-shot semantic aligners directly in the wave domain.
result SIMs achieve up to 90% task accuracy in high SNR regimes, robust to low SNR.
Two semi-supervised manifold alignment methods improve cross-domain classification.
problem Aligning data from multiple sources for better analysis.
method SPUD and MASH methods using graph integration and diffusion.
result SPUD and MASH methods outperform existing methods in cross-domain classification.
LPL optimizes embeddings to align local neighborhoods, improving cross-lingual word alignment.
problem Aligning embeddings across different datasets and languages.
method Locality Preserving Loss (LPL) optimizes model to project embeddings while maintaining local neighborhoods and aligning them.
result LPL-based alignment leads to better and consistent accuracy, especially in small training set settings.
New method separates graph structure from node attributes to recover lost signal.
problem Standard representation learning on attributed graphs merges incompatible metric spaces, leading to geometrically flawed alignment.
method Custom variational autoencoder that separates manifold learning from structural alignment.
result Transforms geometric conflict into interpretable structural descriptor, uncovering connectivity patterns and anomalies.
HSACC improves multi-view clustering of incomplete data.
problem Challenges in clustering incomplete multi-view data.
method Hierarchical Semantic Alignment and Cooperative Completion framework.
result HSACC outperforms state-of-the-art methods on benchmark datasets.
New measures link neural representation geometry to decoding ability.
problem Understanding how neural representations relate to decoding ability.
method Showed that popular similarity measures can be interpreted from a decoding perspective.
result Proved that measures like CKA and CCA quantify alignment between optimal linear readouts.
LLM safety alignment explained as divergence estimation.
problem Aligning large language models to avoid harmful outputs.
method Presented a theoretical framework showing alignment methods as divergence estimators.
result KLDO method improves safety alignment using compliance-refusal datasets.
This paper focuses on the problem of unsupervised alignment of hierarchical data such as ontologies or lexical databases. This is a problem that appears across areas, from natural language processing to bioinformatics, and is typically solved by appeal to outside knowledge bases and label-textual similarity. In contras…
Study Einstein metrics on aligned homogeneous spaces with maximal third Betti number.
problem Existence and classification of Einstein metrics on specific homogeneous spaces.
method Analysis of isotropy representation and computation of Ricci curvature.
result Computation of Ricci curvature formulas for aligned homogeneous spaces.
This paper improves cross-domain learning using random forests for manifold alignment.
problem Improving cross-domain learning and feature integration.
method Semi-supervised manifold alignment using random forest proximities.
result Random forest proximities enhance downstream classification accuracy.
Enhances thematic investing with stock embeddings from textual data.
problem Challenges in constructing thematic portfolios due to overlapping sector boundaries and evolving market dynamics.
method Introduces THEME, a framework that fine-tunes embeddings using hierarchical contrastive learning, aligning themes and stocks using their hierarchical relationship and incorporating stock returns.
result Theme-aligned portfolios demonstrate compelling performance, significantly outperforming large language models in thematic asset retrieval.
In this paper we introduce the deep kernelized autoencoder, a neural network model that allows an explicit approximation of (i) the mapping from an input space to an arbitrary, user-specified kernel space and (ii) the back-projection from such a kernel space to input space. The proposed method is based on traditional a…
A new method for aligning multiple distributions efficiently.
problem Aligning multiple distributions in a shared latent space.
method Iterative alignment of variational approximations of distribution divergences using invertible alignment maps.
result Our method achieves competitive distribution alignment at low computational cost.
A new multi-view clustering method using deep matrix decomposition and partition alignment.
problem Improving multi-view clustering methods to better utilize data representations and view-specific structures.
method Deep matrix decomposition for partition representations, joint use of partition representations, and alternating optimization.
result Demonstrated effectiveness on six benchmark datasets compared to state-of-the-art methods.
Paper tackles entity matching over multi-source data, optimizing alignment and mitigating negative transfer.
problem Learning effective entity matching models over multi-source large-scale data with relaxed assumptions.
method Proposes a Relaxed Multi-source Large-scale Entity-matching (RMLE) problem and Incentive Compatible Pareto Alignment (ICPA) method.
result Optimized cross-source alignments and mitigated negative transfer, improving entity matching accuracy.
SARA uses similarity to learn rewards robustly and adaptively.
problem Robustness to labeler errors and adaptability to diverse feedback formats.
method Contrastive framework that learns latent representations and computes rewards as similarities.
result Strong performance on offline RL benchmarks and diverse applications.
MetFA aligns source and target domains for cross-device image classification.
problem Learning discriminative class boundaries across different domains.
method Distance metric guided feature alignment (MetFA) for domain-invariant and discriminative feature extraction.
result MetFA outperforms state-of-the-art methods in cross-device image classification.
MALI aligns distinct domains using labeled data.
problem Aligning multi-domain data for machine learning.
method MALI learns manifold structure via diffusion and uses labeled data to guide alignment.
result MALI outperforms state-of-the-art methods across multiple datasets.
The paper proves probabilistic alignment between unseen modalities using contrastive learning.
problem Aligning unseen modalities in unsupervised learning.
method Bayesian approach and direct comparison of contrastive representations.
result Direct comparison of contrastive representations recovers the same likelihood ratio as probabilistic graphical models.
Training-free source selection for LLM families with shared vocabularies
problem Source selection for LLM families with shared vocabularies
method Fisher alignment at vocabulary scale
result Fisher alignment is a cosine between kernel mean embeddings in the joint activation-error space