arXiv subjects

Iker Perez

Publications and source records attributed to Iker Perez.

9 recordsLinked to original sources

arXiv2026

Decoupling Inference from State Updates in Low-Latency Feature Engines via Probabilistic Thinning

Streaming data systems increasingly underpin Machine Learning workflows that maintain large numbers of continuously updated aggregations. In production settings, each incoming event typically triggers read-modify-write operations to persistent storage, making high-frequency state updates a dominant source of latency, contention, and operational cost. In this work, we decouple inference from state persistence in streaming Machine Learning pipelines via probabilistic thinning: every event is scored, but durable state updates are selectively triggered by informative events. Unlike approaches that shed input or state, we show that persistence-path control is achievable without a high-frequency in-memory control plane or cross-worker coordination, relying exclusively on approximate statistics retrieved from disk-backed key-value stores. We model the resulting stochastic processes, derive bounds on filtering rates, and prove that common time-based aggregations remain unbiased under variance-aware formulations, preventing systemic error accumulation. We evaluate the approach in a controlled setting that isolates per-event costs, demonstrating substantial reductions in storage Input/Output and serialization overhead. Across experiments, up to 90% of events are excluded from the persistence path while preserving and in some cases improving downstream utility.

cs.DB↗

arXiv2026

Rethinking XAI Evaluation: A Human-Centered Audit of Shapley Benchmarks in High-Stakes Settings

Shapley values are a cornerstone of explainable AI, yet their proliferation into competing formulations has created a fragmented landscape with little consensus on practical deployment. While theoretical differences are well-documented, evaluation remains reliant on quantitative proxies whose alignment with human utility is unverified. In this work, we use a unified amortized framework to isolate semantic differences between eight Shapley variants under the low-latency constraints of operational risk workflows. We conduct a large-scale empirical evaluation across four risk datasets and a realistic fraud-detection environment involving professional analysts and 3,735 case reviews. Our results reveal a fundamental misalignment: standard quantitative metrics, such as sparsity and faithfulness, are decoupled from human-perceived clarity and decision utility. Furthermore, while no formulation improved objective analyst performance, explanations consistently increased decision confidence, signaling a critical risk of automation bias in high-stakes settings. These findings suggest that current evaluation proxies are insufficient for predicting downstream human impact, and we provide evidence-based guidance for selecting formulations and metrics in operational decision systems.

cs.LG↗

arXiv2024

Towards a Foundation Purchasing Model: Pretrained Generative Autoregression on Transaction Sequences

Machine learning models underpin many modern financial systems for use cases such as fraud detection and churn prediction. Most are based on supervised learning with hand-engineered features, which relies heavily on the availability of labelled data. Large self-supervised generative models have shown tremendous success in natural language processing and computer vision, yet so far they haven't been adapted to multivariate time series of financial transactions. In this paper, we present a generative pretraining method that can be used to obtain contextualised embeddings of financial transactions. Benchmarks on public datasets demonstrate that it outperforms state-of-the-art self-supervised methods on a range of downstream tasks. We additionally perform large-scale pretraining of an embedding model using a corpus of data from 180 issuing banks containing 5.1 billion transactions and apply it to the card fraud detection problem on hold-out datasets. The embedding model significantly improves value detection rate at high precision thresholds and transfers well to out-of-domain distributions.

cs.LG↗

arXiv2024

Locally Differentially Private Embedding Models in Distributed Fraud Prevention Systems

Global financial crime activity is driving demand for machine learning solutions in fraud prevention. However, prevention systems are commonly serviced to financial institutions in isolation, and few provisions exist for data sharing due to fears of unintentional leaks and adversarial attacks. Collaborative learning advances in finance are rare, and it is hard to find real-world insights derived from privacy-preserving data processing systems. In this paper, we present a collaborative deep learning framework for fraud prevention, designed from a privacy standpoint, and awarded at the recent PETs Prize Challenges. We leverage latent embedded representations of varied-length transaction sequences, along with local differential privacy, in order to construct a data release mechanism which can securely inform externally hosted fraud and anomaly detection models. We assess our contribution on two distributed data sets donated by large payment networks, and demonstrate robustness to popular inference-time attacks, along with utility-privacy trade-offs analogous to published work in alternative application domains.

cs.CR↗

arXiv2022

Attribution of Predictive Uncertainties in Classification Models

Predictive uncertainties in classification tasks are often a consequence of model inadequacy or insufficient training data. In popular applications, such as image processing, we are often required to scrutinise these uncertainties by meaningfully attributing them to input features. This helps to improve interpretability assessments. However, there exist few effective frameworks for this purpose. Vanilla forms of popular methods for the provision of saliency masks, such as SHAP or integrated gradients, adapt poorly to target measures of uncertainty. Thus, state-of-the-art tools instead proceed by creating counterfactual or adversarial feature vectors, and assign attributions by direct comparison to original images. In this paper, we present a novel framework that combines path integrals, counterfactual explanations and generative models, in order to procure attributions that contain few observable artefacts or noise. We evidence that this outperforms existing alternatives through quantitative evaluations with popular benchmarking methods and data sets of varying complexity.

cs.LG↗

arXiv2019

Variational inequalities and mean-field approximations for partially observed systems of queueing networks

Queueing networks are systems of theoretical interest that find widespread use in the performance evaluation of interconnected resources. In comparison to counterpart models in genetics or mathematical biology, the stochastic (jump) processes induced by queueing networks have distinctive coupling and synchronization properties. This has prevented the derivation of variational approximations for conditional representations of transient dynamics, which rely on simplifying independence assumptions. Here, we present a model augmentation to a multivariate counting process for interactions across service stations, and we enable the variational evaluation of mean-field measures for partially-observed multi-class networks. We also show that our framework offers an efficient and improved alternative for inference tasks, where existing variational or numerically intensive solutions do not work.

stat.ME↗

arXiv2019

Scalable Bayesian Inference for Population Markov Jump Processes

Bayesian inference for Markov jump processes (MJPs) where available observations relate to either system states or jumps typically relies on data-augmentation Markov Chain Monte Carlo. State-of-the-art developments involve representing MJP paths with auxiliary candidate jump times that are later thinned. However, these algorithms are i) unfeasible in situations involving large or infinite capacity systems and ii) not amenable for all observation types. In this paper we establish and present a general data-augmentation framework for population MJPs based on uniformized representations of the underlying non-stationary jump processes. This leads to multiple novel MCMC samplers which enable exact (in the Monte Carlo sense) inference tasks for model parameters. We show that proposed samplers outperform existing popular approaches, and offer substantial efficiency gains in applications to partially observed stochastic epidemics, immigration processes and predator-prey dynamical systems.

stat.CO↗

arXiv2018

On Bayesian inferential tasks with infinite-state jump processes: efficient data augmentation

Advances in sampling schemes for Markov jump processes have recently enabled multiple inferential tasks. However, in statistical and machine learning applications, we often require that these continuous-time models find support on structured and infinite state spaces. In these cases, exact sampling may only be achieved by often inefficient particle filtering procedures, and rapidly augmenting observed datasets remains a significant challenge. Here, we build on the principles of uniformization and present a tractable framework to address this problem, which greatly improves the efficiency of existing state-of-the-art methods commonly used in small finite-state systems, and further scales their use to infinite-state scenarios. We capitalize on the marginal role of variable subsets in a model hierarchy during the process jumps, and describe an algorithm that relies on measurable mappings between pairs of states and carefully designed sets of synthetic jump observations. The proposed method enables the efficient integration of slice sampling techniques and it can overcome the existing computational bottleneck. We offer evidence by means of experiments addressing inference and clustering tasks on both simulated and real data sets.

stat.CO↗

arXiv2017

Auxiliary Variables for Bayesian Inference in Multi-Class Queueing Networks

Queue networks describe complex stochastic systems of both theoretical and practical interest. They provide the means to assess alterations, diagnose poor performance and evaluate robustness across sets of interconnected resources. In the present paper, we focus on the underlying continuous-time Markov chains induced by these networks, and we present a flexible method for drawing parameter inference in multi-class Markovian cases with switching and different service disciplines. The approach is directed towards the inferential problem with missing data and introduces a slice sampling technique with mappings to the measurable space of task transitions between service stations. The method deals with time and tractability issues, can handle prior system knowledge and overcomes common restrictions on service rates across existing inferential frameworks. Finally, the proposed algorithm is validated on synthetic data and applied to a real data set, obtained from a service delivery tasking tool implemented in two university hospitals.

stat.CO↗