Research connected to &quot;machine learning&quot;

ELLA: Efficient Lifelong Learning for Adapters in Large Language Models

Large Language Models (LLMs) suffer severe catastrophic forgetting when adapted sequentially to new tasks in a continual learning (CL) setting. Existing approaches are fundamentally limited: replay-based methods are impractical and privacy-violating, while strict orthogonality-based methods collapse under scale: each new task is projected onto an orthogonal complement, progressively reducing the residual degrees of freedom and eliminating forward transfer by forbidding overlap in shared representations. In this work, we introduce ELLA, a training framework built on the principle of selective subspace de-correlation. Rather than forbidding all overlap, ELLA explicitly characterizes the structure of past updates and penalizes alignments along their high-energy, task-specific directions, while preserving freedom in the low-energy residual subspaces to enable transfer. Formally, this is realized via a lightweight regularizer on a single aggregated update matrix. We prove this mechanism corresponds to an anisotropic shrinkage operator that bounds interference, yielding a penalty that is both memory- and compute-constant regardless of task sequence length. ELLA requires no data replay, n

Theoretical Interpretation of Learned Step Size in Deep-Unfolded Gradient Descent

Deep unfolding is a promising deep-learning technique in which an iterative algorithm is unrolled to a deep network architecture with trainable parameters. In the case of gradient descent algorithms, as a result of the training process, one often observes the acceleration of the convergence speed with learned non-constant step size parameters whose behavior is not intuitive nor interpretable from conventional theory. In this paper, we provide a theoretical interpretation of the learned step size of deep-unfolded gradient descent (DUGD). We first prove that the training process of DUGD reduces not only the mean squared error loss but also the spectral radius related to the convergence rate. Next, we show that minimizing the upper bound of the spectral radius naturally leads to the Chebyshev step which is a sequence of the step size based on Chebyshev polynomials. The numerical experiments confirm that the Chebyshev steps qualitatively reproduce the learned step size parameters in DUGD, which provides a plausible interpretation of the learned parameters. Additionally, we show that the Chebyshev steps achieve the lower bound of the convergence rate for the first-order method in a spec

preprint2013arXiv

Learning Topic Models and Latent Bayesian Networks Under Expansion Constraints

Unsupervised estimation of latent variable models is a fundamental problem central to numerous applications of machine learning and statistics. This work presents a principled approach for estimating broad classes of such models, including probabilistic topic models and latent linear Bayesian networks, using only second-order observed moments. The sufficient conditions for identifiability of these models are primarily based on weak expansion constraints on the topic-word matrix, for topic models, and on the directed acyclic graph, for Bayesian networks. Because no assumptions are made on the distribution among the latent variables, the approach can handle arbitrary correlations among the topics or latent factors. In addition, a tractable learning method via $\ell_1$ optimization is proposed and studied in numerical experiments.

Revisiting Outer Optimization in Adversarial Training

Despite the fundamental distinction between adversarial and natural training (AT and NT), AT methods generally adopt momentum SGD (MSGD) for the outer optimization. This paper aims to analyze this choice by investigating the overlooked role of outer optimization in AT. Our exploratory evaluations reveal that AT induces higher gradient norm and variance compared to NT. This phenomenon hinders the outer optimization in AT since the convergence rate of MSGD is highly dependent on the variance of the gradients. To this end, we propose an optimization method called ENGM which regularizes the contribution of each input example to the average mini-batch gradients. We prove that the convergence rate of ENGM is independent of the variance of the gradients, and thus, it is suitable for AT. We introduce a trick to reduce the computational cost of ENGM using empirical observations on the correlation between the norm of gradients w.r.t. the network parameters and input examples. Our extensive evaluations and ablation studies on CIFAR-10, CIFAR-100, and TinyImageNet demonstrate that ENGM and its variants consistently improve the performance of a wide range of AT methods. Furthermore, ENGM alleviates major shortcomings of AT including robust overfitting and high sensitivity to hyperparameter settings.

Recurrent Neural Radio Anomaly Detection

We introduce a powerful recurrent neural network based method for novelty detection to the application of detecting radio anomalies. This approach holds promise in significantly increasing the ability of naive anomaly detection to detect small anomalies in highly complex complexity multi-user radio bands. We demonstrate the efficacy of this approach on a number of common real over the air radio communications bands of interest and quantify detection performance in terms of probability of detection an false alarm rates across a range of interference to band power ratios and compare to baseline methods.

SCHA-VAE: Hierarchical Context Aggregation for Few-Shot Generation

A few-shot generative model should be able to generate data from a novel distribution by only observing a limited set of examples. In few-shot learning the model is trained on data from many sets from distributions sharing some underlying properties such as sets of characters from different alphabets or objects from different categories. We extend current latent variable models for sets to a fully hierarchical approach with an attention-based point to set-level aggregation and call our method SCHA-VAE for Set-Context-Hierarchical-Aggregation Variational Autoencoder. We explore likelihood-based model comparison, iterative data sampling, and adaptation-free out-of-distribution generalization. Our results show that the hierarchical formulation better captures the intrinsic variability within the sets in the small data regime. This work generalizes deep latent variable approaches to few-shot learning, taking a step toward large-scale few-shot generation with a formulation that readily works with current state-of-the-art deep generative models.

A Comparison of Uncertainty Estimation Approaches in Deep Learning Components for Autonomous Vehicle Applications

A key factor for ensuring safety in Autonomous Vehicles (AVs) is to avoid any abnormal behaviors under undesirable and unpredicted circumstances. As AVs increasingly rely on Deep Neural Networks (DNNs) to perform safety-critical tasks, different methods for uncertainty quantification have recently been proposed to measure the inevitable source of errors in data and models. However, uncertainty quantification in DNNs is still a challenging task. These methods require a higher computational load, a higher memory footprint, and introduce extra latency, which can be prohibitive in safety-critical applications. In this paper, we provide a brief and comparative survey of methods for uncertainty quantification in DNNs along with existing metrics to evaluate uncertainty predictions. We are particularly interested in understanding the advantages and downsides of each method for specific AV tasks and types of uncertainty sources.

Exploring the Advantages of Dense-Vector to One-Hot Encoding of Intent Classes in Out-of-Scope Detection Tasks

This work explores the intrinsic limitations of the popular one-hot encoding method in classification of intents when detection of out-of-scope (OOS) inputs is required. Although recent work has shown that there can be significant improvements in OOS detection when the intent classes are represented as dense-vectors based on domain specific knowledge, we argue in this paper that such gains are more likely due to advantages of dense-vector to one-hot encoding methods in representing the complexity of the OOS space. We start by showing how dense-vector encodings can create OOS spaces with much richer topologies than one-hot encoding methods. We then demonstrate empirically, using four standard intent classification datasets, that knowledge-free, randomly generated dense-vector encodings of intent classes can yield massive, over 20% gains over one-hot encodings, and also outperform the previous, domain knowledge-based, SOTA of one of the datasets. We finish by describing a novel algorithm to search for good dense-vector encodings and present initial but promising experimental results of its use.

preprint2021arXiv

Detecting Malicious Accounts in Permissionless Blockchains using Temporal Graph Properties

The temporal nature of modeling accounts as nodes and transactions as directed edges in a directed graph -- for a blockchain, enables us to understand the behavior (malicious or benign) of the accounts. Predictive classification of accounts as malicious or benign could help users of the permissionless blockchain platforms to operate in a secure manner. Motivated by this, we introduce temporal features such as burst and attractiveness on top of several already used graph properties such as the node degree and clustering coefficient. Using identified features, we train various Machine Learning (ML) algorithms and identify the algorithm that performs the best in detecting which accounts are malicious. We then study the behavior of the accounts over different temporal granularities of the dataset before assigning them malicious tags. For Ethereum blockchain, we identify that for the entire dataset - the ExtraTreesClassifier performs the best among supervised ML algorithms. On the other hand, using cosine similarity on top of the results provided by unsupervised ML algorithms such as K-Means on the entire dataset, we were able to detect 554 more suspicious accounts. Further, using behavior change analysis for accounts, we identify 814 unique suspicious accounts across different temporal granularities.

preprint2019arXiv

A Probabilistic Approach for Discovering Daily Human Mobility Patterns with Mobile Data

Discovering human mobility patterns with geo-location data collected from smartphone users has been a hot research topic in recent years. In this paper, we attempt to discover daily mobile patterns based on GPS data. We view this problem from a probabilistic perspective in order to explore more information from the original GPS data compared to other conventional methods. A non-parameter Bayesian modeling method, Infinite Gaussian Mixture Model, is used to estimate the probability density for the daily mobility. Then, we use Kullback-Leibler divergence as the metrics to measure the similarity of different probability distributions. And combining Infinite Gaussian Mixture Model and Kullback-Leibler divergence, we derived an automatic clustering algorithm to discover mobility patterns for each individual user without setting the number of clusters in advance. In the experiments, the effectiveness of our method is validated on the real user data collected from different users. The results show that the IGMM-based algorithm outperforms the GMM-based algorithm. We also test our methods on the dataset with different lengths to discover the minimum data length for discovering mobility pat

Classification using Hyperdimensional Computing: A Review

Hyperdimensional (HD) computing is built upon its unique data type referred to as hypervectors. The dimension of these hypervectors is typically in the range of tens of thousands. Proposed to solve cognitive tasks, HD computing aims at calculating similarity among its data. Data transformation is realized by three operations, including addition, multiplication and permutation. Its ultra-wide data representation introduces redundancy against noise. Since information is evenly distributed over every bit of the hypervectors, HD computing is inherently robust. Additionally, due to the nature of those three operations, HD computing leads to fast learning ability, high energy efficiency and acceptable accuracy in learning and classification tasks. This paper introduces the background of HD computing, and reviews the data representation, data transformation, and similarity measurement. The orthogonality in high dimensions presents opportunities for flexible computing. To balance the tradeoff between accuracy and efficiency, strategies include but are not limited to encoding, retraining, binarization and hardware acceleration. Evaluations indicate that HD computing shows great potential in

Benefits of depth in neural networks

For any positive integer $k$, there exist neural networks with $Θ(k^3)$ layers, $Θ(1)$ nodes per layer, and $Θ(1)$ distinct parameters which can not be approximated by networks with $\mathcal{O}(k)$ layers unless they are exponentially large --- they must possess $Ω(2^k)$ nodes. This result is proved here for a class of nodes termed "semi-algebraic gates" which includes the common choices of ReLU, maximum, indicator, and piecewise polynomial functions, therefore establishing benefits of depth against not just standard networks with ReLU gates, but also convolutional networks with ReLU and maximization gates, sum-product networks, and boosted decision trees (in this last case with a stronger separation: $Ω(2^{k^3})$ total tree nodes are required).

Iterative Label Improvement: Robust Training by Confidence Based Filtering and Dataset Partitioning

State-of-the-art, high capacity deep neural networks not only require large amounts of labelled training data, they are also highly susceptible to label errors in this data, typically resulting in large efforts and costs and therefore limiting the applicability of deep learning. To alleviate this issue, we propose a novel meta training and labelling scheme that is able to use inexpensive unlabelled data by taking advantage of the generalization power of deep neural networks. We show experimentally that by solely relying on one network architecture and our proposed scheme of iterative training and prediction steps, both label quality and resulting model accuracy can be improved significantly. Our method achieves state-of-the-art results, while being architecture agnostic and therefore broadly applicable. Compared to other methods dealing with erroneous labels, our approach does neither require another network to be trained, nor does it necessarily need an additional, highly accurate reference label set. Instead of removing samples from a labelled set, our technique uses additional sensor data without the need for manual labelling. Furthermore, our approach can be used for semi-super

Mitigating Out-of-Distribution Data Density Overestimation in Energy-Based Models

Deep energy-based models (EBMs), which use deep neural networks (DNNs) as energy functions, are receiving increasing attention due to their ability to learn complex distributions. To train deep EBMs, the maximum likelihood estimation (MLE) with short-run Langevin Monte Carlo (LMC) is often used. While the MLE with short-run LMC is computationally efficient compared to an MLE with full Markov Chain Monte Carlo (MCMC), it often assigns high density to out-of-distribution (OOD) data. To address this issue, here we systematically investigate why the MLE with short-run LMC can converge to EBMs with wrong density estimates, and reveal that the heuristic modifications to LMC introduced by previous works were the main problem. We then propose a Uniform Support Partitioning (USP) scheme that optimizes a set of points to evenly partition the support of the EBM and then uses the resulting points to approximate the EBM-MLE loss gradient. We empirically demonstrate that USP avoids the pitfalls of short-run LMC, leading to significantly improved OOD data detection performance on Fashion-MNIST.

Low-rank features based double transformation matrices learning for image classification

Linear regression is a supervised method that has been widely used in classification tasks. In order to apply linear regression to classification tasks, a technique for relaxing regression targets was proposed. However, methods based on this technique ignore the pressure on a single transformation matrix due to the complex information contained in the data. A single transformation matrix in this case is too strict to provide a flexible projection, thus it is necessary to adopt relaxation on transformation matrix. This paper proposes a double transformation matrices learning method based on latent low-rank feature extraction. The core idea is to use double transformation matrices for relaxation, and jointly projecting the learned principal and salient features from two directions into the label space, which can share the pressure of a single transformation matrix. Firstly, the low-rank features are learned by the latent low rank representation (LatLRR) method which processes the original data from two directions. In this process, sparse noise is also separated, which alleviates its interference on projection learning to some extent. Then, two transformation matrices are introduced to process the two features separately, and the information useful for the classification is extracted. Finally, the two transformation matrices can be easily obtained by alternate optimization methods. Through such processing, even when a large amount of redundant information is contained in samples, our method can also obtain projection results that are easy to classify. Experiments on multiple data sets demonstrate the effectiveness of our approach for classification, especially for complex scenarios.

LLM-Driven Performance-Space Augmentation for Meta-Learning-Based Algorithm Selection

Meta-learning for algorithm selection relies on a meta-dataset in which each row corresponds to a supervised learning dataset described by meta-features and labelled with a target value that is associated with algorithm choice (typically, some function of algorithm performance). A persistent limitation is that the number of curated real-world datasets is small, resulting in sparse meta-datasets that constrain meta-learner generalisation. In this paper, we address this problem by augmenting the meta-dataset with synthetic regression datasets produced via a large language model (LLM), with generation steered toward target regions of a low-dimensionality performance space. In our experiments, we adopt a two-dimensional geometric setting defined by the cross-validated $R^2$ scores of two anchor algorithms, known as landmarkers. We compare two augmentation strategies: (1) uniform sampling, which distributes synthetic datasets across the performance space; and (2) margin-based sampling, which concentrates them near the decision boundary where landmarker preference is most ambiguous. Across 42 real-world UCI regression datasets and 730 synthetic datasets, both strategies substantially improve meta-learner performance over the unaugmented baseline under regression and multi-label evaluation formulations. However, uniform augmentation consistently outperforms margin-based augmentation, achieving a 17.47% relative reduction in Hamming loss, a 100.41% relative improvement in subset accuracy, and a +6.09% relative gain in pooled out-of-fold $R^2$. These results lead us to postulate a central thesis: the performance of algorithms resides on a low-dimensional performance manifold, whose reconstruction bias may be minimised by user-guided LLMs that seek to maximise uniform $ε$-cover, and consequently, lead to improved meta-learning for algorithm selection.

Diversity in Large Language Models under Supervised Fine-Tuning

Supervised Fine-Tuning (SFT) is essential for aligning Large Language Models (LLMs) with user intent, yet it is believed to suppress generative diversity. Although this reduction is frequently referenced, formal empirical testing of the phenomenon remains limited. The expressiveness of LLMs by itself was addressed by multiple prior methods. Their varying perspectives suggest that deeper investigation could yield further improvements. In this study, we attribute the decline to two primary drivers: the neglect of low-frequency patterns within fine-tuning datasets and the forgetting of preexisting knowledge. Motivated by our theoretical analysis, we develop Tempered Focal (TOFU) loss, a novel objective that addresses both stated challenges simultaneously. Our extensive evaluation confirms at scale that generation breadth narrows after SFT and strengthens the hypothesis explaining this effect. Across multiple models and benchmarks, we demonstrate that TOFU enhances output diversity while preserving high response quality, offering a principled approach to SFT.

Pre-training Molecular Graph Representation with 3D Geometry

Molecular graph representation learning is a fundamental problem in modern drug and material discovery. Molecular graphs are typically modeled by their 2D topological structures, but it has been recently discovered that 3D geometric information plays a more vital role in predicting molecular functionalities. However, the lack of 3D information in real-world scenarios has significantly impeded the learning of geometric graph representation. To cope with this challenge, we propose the Graph Multi-View Pre-training (GraphMVP) framework where self-supervised learning (SSL) is performed by leveraging the correspondence and consistency between 2D topological structures and 3D geometric views. GraphMVP effectively learns a 2D molecular graph encoder that is enhanced by richer and more discriminative 3D geometry. We further provide theoretical insights to justify the effectiveness of GraphMVP. Finally, comprehensive experiments show that GraphMVP can consistently outperform existing graph SSL methods.

preprint2021arXiv

Time Adaptive Gaussian Model

Multivariate time series analysis is becoming an integral part of data analysis pipelines. Understanding the individual time point connections between covariates as well as how these connections change in time is non-trivial. To this aim, we propose a novel method that leverages on Hidden Markov Models and Gaussian Graphical Models -- Time Adaptive Gaussian Model (TAGM). Our model is a generalization of state-of-the-art methods for the inference of temporal graphical models, its formulation leverages on both aspects of these models providing better results than current methods. In particular,it performs pattern recognition by clustering data points in time; and, it finds probabilistic (and possibly causal) relationships among the observed variables. Compared to current methods for temporal network inference, it reduces the basic assumptions while still showing good inference performances.

A Data-Augmentation Is Worth A Thousand Samples: Exact Quantification From Analytical Augmented Sample Moments

Data-Augmentation (DA) is known to improve performance across tasks and datasets. We propose a method to theoretically analyze the effect of DA and study questions such as: how many augmented samples are needed to correctly estimate the information encoded by that DA? How does the augmentation policy impact the final parameters of a model? We derive several quantities in close-form, such as the expectation and variance of an image, loss, and model's output under a given DA distribution. Those derivations open new avenues to quantify the benefits and limitations of DA. For example, we show that common DAs require tens of thousands of samples for the loss at hand to be correctly estimated and for the model training to converge. We show that for a training loss to be stable under DA sampling, the model's saliency map (gradient of the loss with respect to the model's input) must align with the smallest eigenvector of the sample variance under the considered DA augmentation, hinting at a possible explanation on why models tend to shift their focus from edges to textures.

Boolean Decision Rules for Reinforcement Learning Policy Summarisation

Explainability of Reinforcement Learning (RL) policies remains a challenging research problem, particularly when considering RL in a safety context. Understanding the decisions and intentions of an RL policy offer avenues to incorporate safety into the policy by limiting undesirable actions. We propose the use of a Boolean Decision Rules model to create a post-hoc rule-based summary of an agent's policy. We evaluate our proposed approach using a DQN agent trained on an implementation of a lava gridworld and show that, given a hand-crafted feature representation of this gridworld, simple generalised rules can be created, giving a post-hoc explainable summary of the agent's policy. We discuss possible avenues to introduce safety into a RL agent's policy by using rules generated by this rule-based model as constraints imposed on the agent's policy, as well as discuss how creating simple rule summaries of an agent's policy may help in the debugging process of RL agents.

A Mixed-Integer Programming Approach to Training Dense Neural Networks

Artificial Neural Networks (ANNs) are prevalent machine learning models that are applied across various real-world classification tasks. However, training ANNs is time-consuming and the resulting models take a lot of memory to deploy. In order to train more parsimonious ANNs, we propose a novel mixed-integer programming (MIP) formulation for training fully-connected ANNs. Our formulations can account for both binary and rectified linear unit (ReLU) activations, and for the use of a log-likelihood loss. We present numerical experiments comparing our MIP-based methods against existing approaches and show that we are able to achieve competitive out-of-sample performance with more parsimonious models.

On the Matrix-Free Generation of Adversarial Perturbations for Black-Box Attacks

In general, adversarial perturbations superimposed on inputs are realistic threats for a deep neural network (DNN). In this paper, we propose a practical generation method of such adversarial perturbation to be applied to black-box attacks that demand access to an input-output relationship only. Thus, the attackers generate such perturbation without invoking inner functions and/or accessing the inner states of a DNN. Unlike the earlier studies, the algorithm to generate the perturbation presented in this study requires much fewer query trials. Moreover, to show the effectiveness of the adversarial perturbation extracted, we experiment with a DNN for semantic segmentation. The result shows that the network is easily deceived with the perturbation generated than using uniformly distributed random noise with the same magnitude.

Online Learning to Rank with Top-k Feedback

We consider two settings of online learning to rank where feedback is restricted to top ranked items. The problem is cast as an online game between a learner and sequence of users, over $T$ rounds. In both settings, the learners objective is to present ranked list of items to the users. The learner's performance is judged on the entire ranked list and true relevances of the items. However, the learner receives highly restricted feedback at end of each round, in form of relevances of only the top $k$ ranked items, where $k \ll m$. The first setting is \emph{non-contextual}, where the list of items to be ranked is fixed. The second setting is \emph{contextual}, where lists of items vary, in form of traditional query-document lists. No stochastic assumption is made on the generation process of relevances of items and contexts. We provide efficient ranking strategies for both the settings. The strategies achieve $O(T^{2/3})$ regret, where regret is based on popular ranking measures in first setting and ranking surrogates in second setting. We also provide impossibility results for certain ranking measures and a certain class of surrogates, when feedback is restricted to the top ran

Truncated Linear Regression in High Dimensions

As in standard linear regression, in truncated linear regression, we are given access to observations $(A_i, y_i)_i$ whose dependent variable equals $y_i= A_i^{\rm T} \cdot x^* + η_i$, where $x^*$ is some fixed unknown vector of interest and $η_i$ is independent noise; except we are only given an observation if its dependent variable $y_i$ lies in some "truncation set" $S \subset \mathbb{R}$. The goal is to recover $x^*$ under some favorable conditions on the $A_i$'s and the noise distribution. We prove that there exists a computationally and statistically efficient method for recovering $k$-sparse $n$-dimensional vectors $x^*$ from $m$ truncated samples, which attains an optimal $\ell_2$ reconstruction error of $O(\sqrt{(k \log n)/m})$. As a corollary, our guarantees imply a computationally efficient and information-theoretically optimal algorithm for compressed sensing with truncation, which may arise from measurement saturation effects. Our result follows from a statistical and computational analysis of the Stochastic Gradient Descent (SGD) algorithm for solving a natural adaptation of the LASSO optimization problem that accommodates truncation. This generalizes the

Long Range Language Modeling via Gated State Spaces

State space models have shown to be effective at modeling long range dependencies, specially on sequence classification tasks. In this work we focus on autoregressive sequence modeling over English books, Github source code and ArXiv mathematics articles. Based on recent developments around the effectiveness of gated activation functions, we propose a new layer named Gated State Space (GSS) and show that it trains significantly faster than the diagonal version of S4 (i.e. DSS) on TPUs, is fairly competitive with several well-tuned Transformer-based baselines and exhibits zero-shot generalization to longer inputs while being straightforward to implement. Finally, we show that leveraging self-attention to model local dependencies improves the performance of GSS even further.

preprint2021arXiv

Active Learning with Multifidelity Modeling for Efficient Rare Event Simulation

While multifidelity modeling provides a cost-effective way to conduct uncertainty quantification with computationally expensive models, much greater efficiency can be achieved by adaptively deciding the number of required high-fidelity (HF) simulations, depending on the type and complexity of the problem and the desired accuracy in the results. We propose a framework for active learning with multifidelity modeling emphasizing the efficient estimation of rare events. Our framework works by fusing a low-fidelity (LF) prediction with an HF-inferred correction, filtering the corrected LF prediction to decide whether to call the high-fidelity model, and for enhanced subsequent accuracy, adapting the correction for the LF prediction after every HF model call. The framework does not make any assumptions as to the LF model type or its correlations with the HF model. In addition, for improved robustness when estimating smaller failure probabilities, we propose using dynamic active learning functions that decide when to call the HF model. We demonstrate our framework using several academic case studies and two finite element (FE) model case studies: estimating Navier-Stokes velocities using the Stokes approximation and estimating stresses in a transversely isotropic model subjected to displacements via a coarsely meshed isotropic model. Across these case studies, not only did the proposed framework estimate the failure probabilities accurately, but compared with either Monte Carlo or a standard variance reduction method, it also required only a small fraction of the calls to the HF model.

Sequential Consensus for Multi-Agent LLM Debates: A Wald-SPRT compute governor with calibration-based failure detection

Multi-agent LLM debate improves factuality and reasoning, but most recipes pick a fixed round count, over-spending on easy items and under-spending on hard ones. We adapt Wald's Sequential Probability Ratio Test (SPRT) as a plug-in compute governor for LLM debates. After each round, an LLM judge emits a [0,1] consensus score on the latest agent positions; a Wald monitor accumulates the log-likelihood ratio of "useful convergence" vs "not yet useful" under a Beta likelihood family, and stops when either boundary is crossed or returns a capped best-effort outcome at R_max. Under i.i.d. assumptions the rule inherits SPRT type-I/type-II error guarantees; in deployment the calibration itself is the more important object, since it estimates whether the judge score actually separates useful from unhelpful convergence in a given domain. We evaluate two tracks: (i) a Monte-Carlo study under calibrated Beta models characterising working curves, error rates, capping behaviour, and sensitivity; and (ii) a real-LLM evaluation on 200 attempted MMLU and 200 attempted GSM8K items with three heterogeneous agents (gpt-5, claude-opus-4-6, gemini-2.5-pro) and a claude-opus-4-6 judge, using disjoint 40-item calibration subsets. On GSM8K the rule stops in 1.01 average rounds (4.06 LLM calls) at 97.0% accuracy vs 99.0% for fixed-5 debate at 15 calls: a 3.7x call reduction at -2pp accuracy. On MMLU the calibrated KL collapses to about 0 and the rule caps on 99.5% of items at 2.1x cost. The takeaway is not that SPRT makes debate more accurate, but that a classical sequential test serves as a cheap compute-control and failure-detection layer for multi-agent LLM systems.

LLM-ADAM: A Generalizable LLM Agent Framework for Pre-Print Anomaly Detection in Additive Manufacturing

Additive manufacturing (AM) continues to transform modern manufacturing by enabling flexible, on-demand production of complex geometries across diverse industries. Fused filament fabrication (FFF) has extended AM to laboratories, classrooms, and small production environments, but this accessibility shifts process-planning responsibility to users who may lack manufacturing expertise. A syntactically valid slicer profile can still encode thermally or geometrically harmful settings, and subtle G-code edits can alter extrusion, cooling, or adhesion before a print begins. Pre-print G-code screening catches accidental or adversarial machine-program errors before material or machine time is wasted. This paper proposes LLM-ADAM as a generalizable LLM framework for pre-print anomaly detection in AM. The framework decomposes the task into three roles: Extractor-LLM maps a G-code file to a structured process-parameter schema; Reference-LLM converts printer and material documentation into aligned operating ranges; and Judge-LLM interprets a deterministic deviation table and G-code evidence to decide whether a part is non-defective or belongs to an anomaly class. Printers, materials, and LLM backbones are interchangeable test conditions, not fixed assumptions. We evaluate the framework on an N=200 FFF G-code corpus spanning two desktop printer families, two materials, and five classes including non-defective, under-extrusion, over-extrusion, warping, and stringing. The best framework configuration reaches 87.5% accuracy, compared with 59.5% for the strongest engineered single-LLM baseline. The results show that structured decomposition, rather than backbone strength alone, is the dominant source of improvement, with defect classes identified at or near ceiling for leading configurations while residual errors concentrate on conservative false alarms for non-defective samples.

Gradient Based Activations for Accurate Bias-Free Learning

Bias mitigation in machine learning models is imperative, yet challenging. While several approaches have been proposed, one view towards mitigating bias is through adversarial learning. A discriminator is used to identify the bias attributes such as gender, age or race in question. This discriminator is used adversarially to ensure that it cannot distinguish the bias attributes. The main drawback in such a model is that it directly introduces a trade-off with accuracy as the features that the discriminator deems to be sensitive for discrimination of bias could be correlated with classification. In this work we solve the problem. We show that a biased discriminator can actually be used to improve this bias-accuracy tradeoff. Specifically, this is achieved by using a feature masking approach using the discriminator's gradients. We ensure that the features favoured for the bias discrimination are de-emphasized and the unbiased features are enhanced during classification. We show that this simple approach works well to reduce bias as well as improve accuracy significantly. We evaluate the proposed model on standard benchmarks. We improve the accuracy of the adversarial methods while maintaining or even improving the unbiasness and also outperform several other recent methods.

Realistic Evaluation of Transductive Few-Shot Learning

Transductive inference is widely used in few-shot learning, as it leverages the statistics of the unlabeled query set of a few-shot task, typically yielding substantially better performances than its inductive counterpart. The current few-shot benchmarks use perfectly class-balanced tasks at inference. We argue that such an artificial regularity is unrealistic, as it assumes that the marginal label probability of the testing samples is known and fixed to the uniform distribution. In fact, in realistic scenarios, the unlabeled query sets come with arbitrary and unknown label marginals. We introduce and study the effect of arbitrary class distributions within the query sets of few-shot tasks at inference, removing the class-balance artefact. Specifically, we model the marginal probabilities of the classes as Dirichlet-distributed random variables, which yields a principled and realistic sampling within the simplex. This leverages the current few-shot benchmarks, building testing tasks with arbitrary class distributions. We evaluate experimentally state-of-the-art transductive methods over 3 widely used data sets, and observe, surprisingly, substantial performance drops, even below inductive methods in some cases. Furthermore, we propose a generalization of the mutual-information loss, based on $α$-divergences, which can handle effectively class-distribution variations. Empirically, we show that our transductive $α$-divergence optimization outperforms state-of-the-art methods across several data sets, models and few-shot settings. Our code is publicly available at https://github.com/oveilleux/Realistic_Transductive_Few_Shot.