Who Guards the Guardians? The Challenges of Evaluating Identifiability of Learned Representations

• URL: http://arxiv.org/abs/2602.24278v1
• Date: Fri, 27 Feb 2026 18:50:13 GMT
• Title: Who Guards the Guardians? The Challenges of Evaluating Identifiability of Learned Representations
• Abstract summary: We show that each metric implicitly encodes assumptions about both the data-generating process (DGP) and the encoder.<n>When these assumptions are violated, metrics become misspecified and can produce systematic false positives and false negatives.<n>We introduce a taxonomy separating DGP assumptions from encoder geometry, use it to characterise the validity domains of existing metrics, and release an evaluation suite for reproducible stress testing and comparison.
• Score: 25.62598569130843
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Identifiability in representation learning is commonly evaluated using standard metrics (e.g., MCC, DCI, R^2) on synthetic benchmarks with known ground-truth factors. These metrics are assumed to reflect recovery up to the equivalence class guaranteed by identifiability theory. We show that this assumption holds only under specific structural conditions: each metric implicitly encodes assumptions about both the data-generating process (DGP) and the encoder. When these assumptions are violated, metrics become misspecified and can produce systematic false positives and false negatives. Such failures occur both within classical identifiability regimes and in post-hoc settings where identifiability is most needed. We introduce a taxonomy separating DGP assumptions from encoder geometry, use it to characterise the validity domains of existing metrics, and release an evaluation suite for reproducible stress testing and comparison.

Related papers

• Structuring Semantic Embeddings for Principle Evaluation: A Prototype-Guided Contrastive Learning Approach [58.15427952222424]
  General-purpose text embeddings are widely deployed for such tasks, but broad semantic similarity can place semantically similar yet task-distinct examples.<n>We introduce Prototype-Guided Contrastive Learning (PGCL), a prototype-guided geometric regularization module built on top of frozen text embeddings.
  arXiv  Detail & Related papers  (2026-08-15T13:19:25Z)
• Interpretable Coreference Resolution Evaluation Using Explicit Semantics [35.71215348826291]
  Coreference resolution is evaluated using aggregate statistical metrics such as CoNLL-F1.<n>We introduce a semantically-enhanced evaluation framework for coreference resolution.<n>We show that our framework uncovers systematic weaknesses that remain obscured by aggregate metrics.
  arXiv  Detail & Related papers  (2026-05-11T14:20:16Z)
• K$α$LOS finds Consensus: A Meta-Algorithm for Evaluating Inter-Annotator Agreement in Complex Vision Tasks [4.297070083645049]
  We propose K$$LOS, a unified meta-algorithm that generalizes the "Localization First" principle to dataset quality evaluation.<n>By resolving spatial correspondence before assessing agreement, our framework transforms complex-categorical problems into nominal reliability.
  arXiv  Detail & Related papers  (2026-03-28T08:54:05Z)
• Beyond Accuracy: Introducing a Symbolic-Mechanistic Approach to Interpretable Evaluation [8.169425244574203]
  We argue for a mechanism-aware evaluation that combines task-relevant symbolic rules with interpretability.<n>Standard evaluation shows the memorization achieves 94% field-name accuracy on unseen data, falsely suggesting competence.<n>Our symbolic-stic evaluation reveals this model violates core schema generalization rules, a failure invisible to accuracy metrics.
  arXiv  Detail & Related papers  (2026-03-06T03:50:59Z)
• The Emergence of Lab-Driven Alignment Signatures: A Psychometric Framework for Auditing Latent Bias and Compounding Risk in Generative AI [0.0]
  This paper introduces a novel auditing framework to quantify latent trait estimation under ordinal uncertainty.<n>The research audits nine leading models across dimensions including Optimization Bias, Sycophancy, and Status-Quo Legitimization.
  arXiv  Detail & Related papers  (2026-02-19T06:56:01Z)
• Partial Identification under Missing Data Using Weak Shadow Variables from Pretrained Models [19.367626631086214]
  Estimating population quantities such as mean outcomes from user feedback is fundamental to platform evaluation and social science.<n>Existing approaches typically rely on strong parametric assumptions or bespoke auxiliary variables that may be unavailable in practice.<n>We develop a partial identification framework in which sharp bounds on the estimand are obtained by solving a pair of linear programs.
  arXiv  Detail & Related papers  (2026-02-17T22:18:27Z)
• SCORE: A Semantic Evaluation Framework for Generative Document Parsing [2.5101597298392098]
  Multi-modal generative document parsing systems produce semantically correct yet structurally divergent outputs.<n>Conventional metrics-CER, WER, IoU, or TEDS-misclassify such diversity as error, penalizing valid interpretations and obscuring system behavior.<n>We introduce SCORE, an interpretation-agnostic framework that integrates (i) adjusted edit distance for robust content fidelity, (ii) token-level diagnostics to distinguish hallucinations from omissions, (iii) table evaluation with spatial tolerance and semantic alignment, and (iv) hierarchy-aware consistency checks.
  arXiv  Detail & Related papers  (2025-09-16T16:06:19Z)
• Performance Estimation in Binary Classification Using Calibrated Confidence [0.5399800035598186]
  We present CBPE, a novel method that can estimate any binary classification metric defined using the confusion matrix.<n>CBPE is shown to produce estimates that come with strong theoretical guarantees and valid confidence intervals.
  arXiv  Detail & Related papers  (2025-05-08T14:34:44Z)
• SEOE: A Scalable and Reliable Semantic Evaluation Framework for Open Domain Event Detection [70.23196257213829]
  We propose a scalable and reliable Semantic-level Evaluation framework for Open domain Event detection.<n>Our proposed framework first constructs a scalable evaluation benchmark that currently includes 564 event types covering 7 major domains.<n>We then leverage large language models (LLMs) as automatic evaluation agents to compute a semantic F1-score, incorporating fine-grained definitions of semantically similar labels.
  arXiv  Detail & Related papers  (2025-03-05T09:37:05Z)
• Causal Fair Metric: Bridging Causality, Individual Fairness, and Adversarial Robustness [7.246701762489971]
  Adversarial perturbation, used to identify vulnerabilities in models, and individual fairness, aiming for equitable treatment of similar individuals, both depend on metrics to generate comparable input data instances. Previous attempts to define such joint metrics often lack general assumptions about data or structural causal models and were unable to reflect counterfactual proximity. This paper introduces a causal fair metric formulated based on causal structures encompassing sensitive attributes and protected causal perturbation.
  arXiv  Detail & Related papers  (2023-10-30T09:53:42Z)
• Disentanglement via Latent Quantization [60.37109712033694]
  In this work, we construct an inductive bias towards encoding to and decoding from an organized latent space. We demonstrate the broad applicability of this approach by adding it to both basic data-re (vanilla autoencoder) and latent-reconstructing (InfoGAN) generative models.
  arXiv  Detail & Related papers  (2023-05-28T06:30:29Z)
• Parametric Classification for Generalized Category Discovery: A Baseline Study [70.73212959385387]
  Generalized Category Discovery (GCD) aims to discover novel categories in unlabelled datasets using knowledge learned from labelled samples. We investigate the failure of parametric classifiers, verify the effectiveness of previous design choices when high-quality supervision is available, and identify unreliable pseudo-labels as a key problem. We propose a simple yet effective parametric classification method that benefits from entropy regularisation, achieves state-of-the-art performance on multiple GCD benchmarks and shows strong robustness to unknown class numbers.
  arXiv  Detail & Related papers  (2022-11-21T18:47:11Z)
• Learning Hidden Markov Models When the Locations of Missing Observations are Unknown [54.40592050737724]
  We consider the general problem of learning an HMM from data with unknown missing observation locations. We provide reconstruction algorithms that do not require any assumptions about the structure of the underlying chain. We show that under proper specifications one can reconstruct the process dynamics as well as if the missing observations positions were known.
  arXiv  Detail & Related papers  (2022-03-12T22:40:43Z)
• Self-Certifying Classification by Linearized Deep Assignment [65.0100925582087]
  We propose a novel class of deep predictors for classifying metric data on graphs within PAC-Bayes risk certification paradigm. Building on the recent PAC-Bayes literature and data-dependent priors, this approach enables learning posterior distributions on the hypothesis space.
  arXiv  Detail & Related papers  (2022-01-26T19:59:14Z)
• Posthoc Verification and the Fallibility of the Ground Truth [10.427125361534966]
  We conduct a systematic posthoc verification experiment on the entity linking (EL) task. Compared to pre-annotation evaluation, state-of-the-art EL models performed extremely well according to the posthoc evaluation methodology. Surprisingly, we find predictions from EL models had a similar or higher verification rate than the ground truth.
  arXiv  Detail & Related papers  (2021-06-02T17:57:09Z)
• Certified Robustness to Label-Flipping Attacks via Randomized Smoothing [105.91827623768724]
  Machine learning algorithms are susceptible to data poisoning attacks. We present a unifying view of randomized smoothing over arbitrary functions. We propose a new strategy for building classifiers that are pointwise-certifiably robust to general data poisoning attacks.
  arXiv  Detail & Related papers  (2020-02-07T21:28:30Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.