Sección

Investigación

Papers y preprints de los feeds de investigación que seguimos.

6415documentos

SOAR: Social and Opinion-Aware Recommendations

One of the enduring challenges in online platforms is recommendation. Both communities and advertisers face difficulties in analyzing complex user behavior, identifying habits and evolving preferences, and providing timely recommendations that are relevant to user interests. In particular, point of interest recommendation poses the additional challenge of identifying the factors that influence user behavior, capturing user tendencies, and generating accurate, personalized recommendations for each individual. In this work, we propose SOAR, a Social and Opinion Aware Recommendation model that incorporates both the user's friendship network and textual reviews to assess whether a location is suitable for a specific user, with a particular emphasis on the sentiment of shared reviews. Experimental results in real-world datasets demonstrate that incorporating reviews into the recommendation process improves the precision of the recommendation.

arXiv AI Papershace 23 días

WRDO: Weighted Refusal Direction Optimization in Aligned Language Models

Aligned large language models are trained to comply with benign requests while refusing harmful ones. However, recent work shows that this refusal behavior is fragile and can often be disrupted by simple activation- or weight-space interventions. In particular, refusal has been shown to be largely mediated by a single residual-stream direction, typically computed as a difference-of-means vector between harmful and harmless prompts. However, this standard construction aggregates heterogeneous harmful prompts into one uniform average, potentially conflating distinct refusal-related phenomena. In this work, we study whether refusal interventions can be strengthened by learning how to combine prompt-specific refusal directions. We propose Weighted Refusal Direction Optimization (WRDO), a procedure that constructs a composite refusal direction from a bank of per-prompt directions using learned aggregation weights. This approach allows different harmful prompts to contribute unequally to the final intervention direction. We evaluate WRDO against prior single-direction methods on three open-weight language models using three guard judges and HarmBench as an external evaluator. We further measure the effect of the intervention on general model capability through standard utility benchmarks. Finally, we investigate a defense-oriented setting in which the same optimization machinery is used in reverse to restore refusal behavior in an abliterated model checkpoint. Our results show that weighted aggregation provides a more effective and flexible mechanism for manipulating refusal representations.

arXiv AI Papershace 23 días

Hyperspherical Simplex Representations for Backward-Compatible Retrieval

Training modern AI models has become increasingly expensive, and updating deployed models can alter their internal feature representations, affecting applications built upon them. In retrieval systems, such updates often require complete gallery re-indexing, a computationally expensive process at scale. Existing backward-compatibility methods enable direct comparison of representations from updated and earlier models, avoiding re-indexing. However, they typically require compatibility-specific training, learned mapping functions, or architectural modifications, and often depend explicitly on the previous model. We propose hyperspherical simplex representations derived from softmax outputs and logits for backward-compatible retrieval across independently trained models. They leverage the simplex structure induced by softmax cross-entropy training with one-hot labels as a common geometric reference across model updates. Since this reference depends on the number of classes, adding new ones expands the simplex, misaligning existing class prototypes with their earlier counterparts. We therefore derive a deterministic linear transformation to restore their alignment. Our theoretical analysis of the formal backward-compatibility definition identifies two geometric conditions under which its inequalities hold in expectation, assuming von Mises--Fisher class distributions with increasing concentration across updates. Aligned regular-simplex representations satisfy these conditions, providing a theoretical basis for the proposed representations under the stated assumptions. Without using additional compatibility-specific losses, learned mappings, or architectural modifications, the proposed representations achieve competitive results on standard compatibility benchmarks, including expanded class sets, more expressive architectures, and independently trained models from public repositories.

arXiv AI Papershace 23 días

PACE: Periodic and Channel-Aligned Encoding of Four-Bit Recurrent States

Linear-attention language models keep their context in a recurrent state of fixed size, where an inference system stores one copy of this state for every sequence it decodes at the same time. Storing the state at lower precision therefore lets more sequences share an accelerator. Existing state quantizers round the state after every generated token and report that four bits lose accuracy unless the bit width is raised or the model is fine-tuned.We introduce PACE, a post-hoc four-bit state quantizer. PACE rounds the state only once every few generated tokens, as chunked kernels already do once per chunk during prefill, and keeps the tokens generated in between in a small buffer. Rounding less often reduces the error that accumulates in the state. Additionally, we find that the large entries of the gated-delta state gather in columns, because some channels of the update written at each token remain large across many tokens. PACE therefore shares each quantization scale within one column. It also keeps the heads that retain rounding error longest at eight bits. On four hybrid models from 4B to 35B parameters, PACE with mixed four- and eight-bit heads is competitive with eight-bit DAMP on the three larger models at about two thirds of its memory, PACE with four-bit heads throughout scores higher than Q-Mamba at equal memory, and rounding every few tokens preserves the decoding speedup of the buffer.

arXiv AI Papershace 23 días

Stochastic Pruning of the DPO Computation Graph: 1.4x Faster Training and Lower Peak Memory

Direct preference optimization (DPO) incurs training costs in output scoring, sequence execution, and backpropagation, so reducing supervision alone does not necessarily save computation. We introduce a unified framework for stochastic pruning of the DPO computation graph that independently samples response positions for scoring, response prefixes for execution, and preference branches for backpropagation. By making these selections before the corresponding operations, the framework omits computation at each stage of a training step. Our analysis characterizes the objective changes induced by position and sequence pruning and uses inverse-probability weighting to obtain a conditionally unbiased single-branch gradient estimator for the sampled objective. Controlled measurements show that position and sequence pruning primarily reduce peak memory, with most output-memory savings attributable to an implementation that retains full response supervision. Branch pruning yields a 1.426step-time speedup over a concatenated DPO baseline and reduces training time by about 30\% relative to stock DPO across two model families, with slightly lower mean accuracy and benchmark-dependent changes in open-ended scores. These findings relate the resource benefits of stochastic graph pruning to the computation omitted and the resulting changes in the training objective.

arXiv AI Papershace 23 días

Block Diffusion for Longitudinal EHR: Accuracy, Efficiency, and Representation Trade-offs

Autoregressive (AR) models have proved effective for zero-shot prediction from longitudinal electronic health records, but impose strictly sequential generation on trajectories containing temporally clustered and interdependent clinical events. We investigate block-based masked diffusion (BD) as an alternative on MIMIC-IV under matched tokenisation, context length, model capacity, and evaluation conditions. Across block sizes, BD exhibits a clear quality--efficiency trade-off, with small-block configurations retaining much of the predictive performance of AR while enabling substantially faster inference. A representative B{=}8 configuration achieves approximately 1.43faster evaluation. Although generative rollout remains below AR in most settings, frozen linear probes on BD and AR representations achieve near-identical performance across clinical prediction tasks, indicating that BD learns comparably informative clinical representations. Increasing the denoising budget yields diminishing returns, while alternative decoding strategies provide limited gains. Overall, our results show that block diffusion is a viable and computationally attractive approach to longitudinal EHR modelling, with further improvements in iterative decoding offering a promising route toward closing the remaining generative performance gap.

arXiv AI Papershace 23 días

Phase Transitions in Learning Driven by Extremal Data

We study an extremal regression problem, an example of multiple instance learning (MIL), where the predictors are given as multisets, called bags. The bag elements (called instances) possess hidden labels which are pooled to produce an observed bag-level label. Max pooling is common in MIL applications, but gives rise to vanishing gradients. To handle this, modern MIL pipelines approximate the max with a softmax, treating the sharpness as a learnable parameter, denoted λ. When the max pooling model is the true data generating process, learning will be successful when λ is increased during gradient descent, thereby driving the dynamics into regions of parameter space with vanishing gradients. To understand the conditions under which gradient descent breaks down, we study a minimal model and derive a low dimensional projection of the gradient flow in the joint limit of large bags n and high dimension d with α=n/d fixed. The flow has two distinct phase transitions: an intrinsic freezing transition related to the random energy model of disordered systems, and a retrieval transition occurring in modern Hopfield networks. Numerical simulation confirms both the phase diagram and the convergence of the finite-size training gradient to its predicted asymptotic form.

arXiv AI Papershace 23 días

Residual-Based Causal Discovery via Pretrained In-Context Quantile Regression

Residual-based causal discovery assumes that each variable is generated from its parents and an independent noise source. It estimates this noise and tests its independence from candidate causes. For identifiable functions such as additive-noise models (ANMs) and the broader class of location-scale noise models (LSNMs), this principle identifies the causal direction, as only the causal model admits this independence. Existing LSNM methods rely on parametric assumptions to obtain standardized residuals, risking model misspecification. In contrast, the RESIT-framework allows for nonparametric estimation of residuals to iteratively identify sink nodes, but was proposed for ANMs. To close this gap, we propose Loscar, an extension of RESIT to LSNMs that estimates the conditional mean and scale via quantile regression, enabling pretrained tabular foundation models as plug-in estimators. We compare it with a feasible generalized least squares (FGLS) variant, which is consistent given suitable basis functions. Across bivariate and multivariate causal-discovery benchmarks, in-context quantile regression substantially outperforms FGLS with flexible function classes, particularly for multivariate causes.

arXiv AI Papershace 23 días

Are Joint-Embedding Predictive Objectives Effective for Longitudinal EHRs?

Masked language modelling (MLM) is the dominant pretraining objective for transformer encoders applied to longitudinal electronic health records (EHRs). However, reconstructing individual events may overemphasise noisy or weakly informative observations and does not explicitly encourage learning higher-level patient states. Joint-Embedding Predictive Architectures (JEPAs) instead learn by predicting representations in latent space and have shown strong performance in a variety of domains. In this paper, we investigate to what extent JEPA can be adapted to EHR sequence representation. Our empirical investigation using MIMIC-IV shows that although all three of our JEPA formulations train stably without representational collapse, they consistently underperform standard MLM. We find that target size, not target heterogeneity, increases prediction error, yet reducing the target size does not recover downstream performance. This suggests that stable latent prediction alone is insufficient for competitive EHR representation learning and highlights the challenge of defining clinically meaningful latent targets for EHR.

arXiv AI Papershace 23 días

RCSD: Reinforcement and Continual Learning for Domain-Specific Speculative Decoding

Speculative decoding accelerates large language model inference by employing a compact draft model to propose candidate tokens that are verified in parallel by a target model. Conventional training relies on soft-label distillation, which forces the draft model to mimic the target's full output distribution, including high-entropy regions that degrade acceptance rates and increase early rejection risk. We propose RCSD (Reinforcement and Continual Learning for Speculative Decoding), a training framework that aligns draft model optimization with the system-level objective of maximizing acceptance length. RCSD leverages naturally occurring accepted and rejected sequences during speculative inference to construct implicit preference pairs, with masked loss to avoid penalizing target-aligned tokens. Evaluated across multiple benchmarks, RCSD improves average acceptance length by 35% over HASS and achieves 1.35x faster end-to-end inference than HASS on practical workloads, without compromising generation quality.

arXiv AI Papershace 23 días

Geometric knowledge spaces, Attention pairing and Architectures

In this paper, we propose a geometric approach to deriving attention mechanisms and exploring potential new transformer architectures. We begin by formalizing the notions of knowledge worlds and attention pairings, then demonstrate how this framework naturally leads to the architectures of standard attention, causal attention, linear attention, tensor attention, and recurrent neural networks. Finally, we introduce several new geometric spaces that may inspire novel neural network architectures.

arXiv AI Papershace 23 días

pplx-embed-v2-late Technical Report

In this report, we introduce pplx-embed-v2-late, a pair of 0.6B and 9B multilingual and multimodal late-interaction models. Both models achieve state-of-the-art results within their respective size classes across a range of text, image, and visual document retrieval benchmarks. On visual document retrieval tasks, pplx-embed-v2-late-0.6b is competitive with models several times larger than itself. The models are trained via representation distillation from an 18B teacher. Consequently, they share an embedding space that supports asymmetric deployment: documents can be indexed with the 9B model and queried with the low-latency 0.6B model. Both models are available at https://huggingface.co/collections/perplexity-ai/pplx-embed-v2.

arXiv AI Papershace 23 días

Beyond the Vote: When Do LLMs Add Value to Retrieval-Based Tabular Classification?

Retrieval-based in-context learning (ICL) enables large language models (LLMs) to predict tabular labels from similar examples, but it remains unclear whether LLM inference improves on direct voting over those same examples. We address this question by comparing four general-purpose LLMs with distance-weighted nearest-neighbor voting (Distance-KNN) using identical retrieved neighbors across 19 tabular classification datasets. With the number of demonstrations fixed, enlarging the retrieval candidate pool improves accuracy for all predictors, yet Distance-KNN achieves the highest mean accuracy at every pool size. More detailed analyses reveal that LLM gains are concentrated where retrieved neighbors exhibit low label consensus and vary across models. These gains further depend on the average feature-space distance to the test query: under low label consensus, some LLMs outperform Distance-KNN in close and intermediate neighborhoods, but all fall behind on average in the most distant group. Complementary experiments on the number of demonstrations, prompt formulation, retrieval representation, and label-balanced demonstration selection largely preserve the overall findings. These results characterize when LLM inference adds predictive value beyond distance-weighted voting in retrieval-based tabular ICL.

arXiv AI Papershace 23 días

Supervision-Adaptive Curriculum and Guarded Anti-Curriculum Learning for Dense Prediction

Curriculum learning orders samples by difficulty, but difficulty alone does not establish supervision reliability for dense prediction. Easy samples can expose shortcuts, hard samples can be informative, and noisy or ambiguous samples can mislead training. Building on our conference method CurriSeg, we present PaceShift, a stabilize-then-challenge paradigm combining supervision-adaptive curriculum selection with guarded spectral post-training. A learning-dynamics view motivates separating supervision reliability from temporal update value and applying spectral challenges after stabilization. The first phase estimates reliability from available supervision, augmentation consistency, model disagreement, and prediction certainty. Temporal statistics guide sample weighting, while pixel-uncertainty weights reduce the early influence of uncertain regions. The second phase ranks predictions from multiple spectral views using supervision agreement, robustness, reference consistency, and penalties. A pairwise preference objective measured against the frozen Phase-I reference, together with an anchor, refines the prediction. The framework supports full, weak, semi-supervised, and model-derived supervision. Experiments across context-entangled content segmentation (CECS), broader dense-prediction tasks, and foundation-model and cross-modal adaptations show accuracy gains. Targeted diagnostics indicate improved pooled calibration and corruption robustness in the evaluated settings. PaceShift preserves the deployed architecture and adds no inference computation. Code will be publicly released.

arXiv AI Papershace 23 días

Mitigating Query Shortcuts in Chain-of-Thought with Context-Aware Decoding

While Chain-of-Thought (CoT) prompting improves the reasoning performance of large language models (LLMs), recent studies have shown that final answers do not always depend on the generated rationales. In particular, LLMs may rely on query-specific shortcuts rather than the information provided by their reasoning steps. In this paper, we investigate whether context-aware decoding (CAD), an existing contrastive decoding method, can mitigate such shortcut behavior when applied to CoT reasoning. We study this application as Rationale Contrastive Decoding (RCD), which contrasts the next-token distributions conditioned on the query with and without its rationale, amplifying tokens whose likelihood increases due to the rationale. We provide a theoretical interpretation of this decoding rule via a KL-regularized reward-maximization objective. Experiments across five reasoning benchmarks demonstrate that RCD improves answer accuracy over standard CoT. Furthermore, intervention analyses using ground-truth and random rationales reveal that RCD increases the behavioral dependence of final answers on the rationale content. Our findings suggest that contrastive decoding offers a simple, training-free approach to mitigating query shortcuts.

arXiv AI Papershace 23 días

BRIDGE: An Acceptance Gate to Block False Improvements in Model Updates, with a Retrospective Clinical Evaluation

Models are updated repeatedly, and an update that looks better on validation data but not on new data is a false improvement that can misguide clinical decisions. The naive rule accepts every positive validation gain and is optimistic when the same data choose and judge the update. BRIDGE is a model-agnostic acceptance gate that decides each update from predictions and the available labels, with fixed settings. Each fold of the data judges the update chosen on the other folds (cross-fitting) to reduce selection bias; acceptance requires the gain's lower bootstrap bound to exceed a margin. The four settings are (A) 11 new datasets, (B) synthetic tasks, (C) 12 retrospective clinical tasks, and (D) 39 supporting datasets. The false-improvement rate falls from 19.1% (naive) to 1.1% (BRIDGE) in (A), 20.7% to 1.7% (B), 20.0% to 0% (C), and 12.5% to 0.5% (D); every 95% confidence interval of the reduction over datasets or tasks excludes zero (not over the clinical sources). An equal-size random draw from the naive acceptances contains more false improvements in (A)-(E) (post hoc, p<0.001). On the OpenML collection (E) of 14 new datasets, the false-improvement rate falls from 25.6% to 1.0%. BRIDGE removes 97% to 100% of the false improvements that the naive rule accepts, and its mean gain per deployed update is 1.8 to 3.1 times the naive rule's. The 95% upper bound on the false-improvement rate is below 5.5%: formal in (B) for independent tasks (3.56%), and conditional on the real data tested (0.33% to 5.40%).

arXiv AI Papershace 23 días

A unified approach for estimation of unnormalized models via Stein operators

In this paper, we develop a unified framework connecting proper scoring rules and Stein's method for the estimation of unnormalized models. We construct a class of proper scoring rules that encompasses score matching, ratio matching, pseudolikelihood, and noise contrastive estimation. We show that these proper scoring rules naturally induce Stein operators. This connection yields a new construction principle for Stein operators, and enables us to study estimators based on these scoring rules through the geometry associated with the induced Stein operators. Using this geometric perspective, we provide a unified characterization of the asymptotic efficiency of the resulting estimators.

arXiv AI Papershace 23 días

Shortcut-Resistant CAM Distillation for Long-Tailed Recognition

Real-world datasets often follow a long-tailed distribution, making generalization to tail classes difficult. We revisit this problem through the lens of shortcut learning, where models prefer the easiest predictive cues (e.g., background or textures) over object-centric semantics, especially under scarce and biased supervision. We find that this tendency is amplified for tail classes: limited examples often share similar contexts, making non-semantic signals highly correlated and thus tempting shortcuts, whereas head classes with diverse appearances and environments encourage more stable object-focused representations. Motivated by this observation, we propose Shortcut-Resistant CAM Distillation (SRCD), a plug-and-play framework that transfers object-focused explanations from head to tail classes. SRCD operates in the Class Activation Map (CAM) space, where a CAM provides a class-specific spatial evidence map for a prediction. SRCD aggregates CAMs from a small set of head-class candidates into a shortcut-resistant teacher using an energy-model weighting based on coherence and concentration, and distills it to the tail-class CAM. We provide a theoretical analysis that quantifies shortcut reliance as shortcut-region evidence mass in CAM space and shows that SRCD suppresses tail shortcuts. Extensive experiments on long-tailed benchmarks consistently improve strong baselines. The code is available at https://github.com/Haifeng3/SRCD.

arXiv AI Papershace 23 días

Keeping Time vs. Understanding Time: Neither Alone Explains Temporal Reasoning in LLMs

Temporal reasoning is widely assumed to decompose into two modes: temporal arithmetic (date and time calculations), and temporal semantics (understanding temporal and causal relations). This dichotomy shapes how benchmarks are designed, how models are trained, and how progress is measured, yet its empirical validity has never been directly tested. We provide two lines of evidence that challenge it. First, we test whether tasks labelled as arithmetic or semantic by the benchmarks support transferable gains within their purported mode. Finetuning on these subsets yields predominantly dataset-specific gains (+6.8 to +23.1 percentage points), while transfer to a different dataset in the same mode ranges from -3.2 to +1.1 percentage points. Across the two benchmark families used in the transfer analysis, this pattern provides little evidence of coherent, transferable temporal abilities. Second, using cognitive-demand-based model evaluation, we examine which capabilities actually drive performance on these benchmarks. We find that quantitative and logical reasoning, the assumed proxies for temporal arithmetic and temporal semantics, are not the primary drivers; knowledge is. Together, these findings show that current temporal reasoning benchmarks do not cleanly realise the arithmetic-semantics distinction and instead largely reward models for remembering how time typically works.

arXiv AI Papershace 23 días

Asynchronous, Generation-Free ABC: Streaming AMIS Reweighting for Heterogeneous Simulator Workloads on HPC

Approximate Bayesian computation (ABC) enables Bayesian inference for models with intractable likelihoods by estimating parameter posterior distributions from simulated data. Sequential ABC methods improve sampling efficiency by iteratively adapting the proposal distribution toward regions of high posterior probability. However, these methods typically operate in generations, where each generation must wait for its slowest simulation before the next can begin. On a parallel machine, this synchronization barrier leaves workers idle and reduces throughput, particularly when simulation runtimes vary substantially. We introduce a generation-free ABC sampler that updates its proposal when simulation results arrive and sends a new candidate to each available worker. The proposal is based on the best particles evaluated so far. To estimate the posterior, we reuse the full history and reweight particles against a reconstructed mixture of sampling proposals. We show how errors in approximating the mixture of sampling proposals affect the estimated posterior and establish consistency for the smooth-ABC posterior when the approximation becomes asymptotically exact. In our experiments, the method achieves well-calibrated posterior estimates where a pyABC baseline with the same kernel and bandwidth schedule under-covers. From the asynchronous run's timing alone, we predict how much a synchronization barrier would slow the sampler. These predictions agree with measured synchronization penalties ranging from 1.2x to 400x. On a Cellular Potts tissue simulator, the asynchronous sampler completes 2.1-2.4x as many evaluations as pyABC in the same hour, and the discrepancy threshold for its best 100 evaluations is 4.1x smaller. With the starting bandwidth set from prior-predictive discrepancies, it also produces a more concentrated two-parameter posterior with a larger effective sample size than pyABC.