Model fusion research aims to aggregate the knowledge of multiple individual models to enhance performance by combining their weights. In this work, we study the inverse problem: investigating whether model fusion can be used to reduce unwanted knowledge. We investigate the effects of model fusion in three scenarios: the learning of shortcuts, social biases, and memorization of training data in fine-tuned language models. Through experiments covering classification and generation tasks, our analysis highlights that shared knowledge among models is enhanced during model fusion, while unshared knowledge is usually forgotten. Based on this observation, we demonstrate the potential of model fusion as a debiasing tool and showcase its efficacy in addressing privacy concerns associated with language models.

Non-parametric multivariate density estimation faces strong statistical and computational bottlenecks, and the more practical approaches impose near-parametric assumptions on the form of the density functions. In this paper, we leverage recent developments to propose a class of non-parametric models which have very attractive computational and statistical properties. Our approach relies on the simple function space assumption that the conditional distribution of each variable conditioned on the other variables has a non-parametric exponential family form.

This paper contributes some original thinkings on how to assess the quality of a generative model. The new evaluation metric, as defined by the distributional precision and recall statistics (PRD), overcomes a major drawback from prior-arts: that evaluation scores are almost exclusively scalar metrics. The author(s) attributes intuitive explanations to this new metric and experimentally reached a conclusion that it is able to disentangle the quality from the coverage, two critical aspects wrt the quality of a learned synthetic sampler. An efficient algorithm is also described and theoretically justified. The quality of this work seems okay, yet I am prone to a neutral-to-negative rating.

I also strongly approve of the suggestion of highlighting more the experimental results in the main paper. RLSs are well known quantities used in randomized sketching and coreset selection to identify influential samples. Similarly, it is known that sampling **and reweighting** rows of a matrix A according to their RLS produces a sketch that whp approximates the true matrix up to a small multiplicative and additive error. The authors prove that sorting the rows in descending order (by RLS), and deterministically selecting them until the RLS falls under a carefully chosen threshold is also sufficient to obtain a provably accurate sketch. Therefore, they propose deterministic RLS selection as a convenient and provably accurate rule for column selection, with improved interpretability over optimization based alternative such as lasso and elastic net.

This paper introduces Theseus, an algorithm for matching MS/MS spectra to peptide in a D.B. This is a challenging and important task. It is important because MS/MS is currently practically the only common high-throughput method to identify which proteins are present in a sample. It is challenging because the data is analog (intensity vs. m/z graphs) and extremely noisy. This work builds upon an impressive body of work that has been dedicated to this problem.

This paper builds on the work of Platanios et al. (2014, 2016) on estimating the accuracy of a set of classifiers for a given task using only unlabeled data, based on the agreement behavior of the classifiers. The current work uses a probabilistic soft logic (PSL) model to infer the error rates of the classifiers. The paper also extends this approach to the case where we have multiple related classification tasks: for instance, classifying noun phrases with regard to their membership in multiple categories, some of which subsume others and some of which are mutually exclusive. The paper shows how a PSL model can take into account these constraints among the categories, yielding better error rate estimates and higher joint classification accuracy. It is well written and the methodology seems sound.

In this paper the authors examine the intuition that interpretability to be the workhorse in detecting adversarial examples of different kinds. That is, if the humanly interpretable attributes are all the same for two images, then the prediction result should only be different if some non-interpretable neurons behave differently. Other than adversarial examples, this work is also highly related to interpretability and explainability questions for DNNs. The basis of their detection mechanism (AmI) lies in determining the sets of neurons (they call attribute witnesses) that are correspond (one-to-one) to a humanly interpretable attributes (like eyeglasses). That means, if the attribute does not change, the neuron should not give a different output, and the other way around if the feature changes, the neuron should change.

Effective management and early detection of BPSD are crucial to reduce the stress and burden on caregivers and healthcare systems. Despite the advancements in machine learning for dementia prediction, there is a considerable gap in utilizing these methods for BPSD prediction. This study aims to fill this gap by presenting a novel personalized framework for BPSD prediction, utilizing physiological signals from smart wearable devices. Our personalized fine-grained BPSD prediction method accurately predicts BPSD occurrences by extracting individual behavioral patterns, while the generalized models identify diverse patterns and differentiate between various BPSD symptoms. Detailed comparisons between the proposed personalized method and conventional generalized methods reveals substantial improvements across all performance metrics, including a 16.0% increase in AUC. These results demonstrate the potential of our proposed method in advancing dementia care by enabling proactive interventions and improving patient outcomes in real-world scenarios. To the best of our knowledge, this is the first study that leverages physiological signals from smart wearable devices to predict BPSD, marking a significant stride in dementia care research.

Gene expression analysis is a critical method for cancer classification, enabling precise diagnoses through the identification of unique molecular signatures associated with various tumors. Identifying cancer-specific genes from gene expression values enables a more tailored and personalized treatment approach. However, the high dimensionality of mRNA gene expression data poses challenges for analysis and data extraction. This research presents a comprehensive pipeline designed to accurately identify 33 distinct cancer types and their corresponding gene sets. It incorporates a combination of normalization and feature selection techniques to reduce dataset dimensionality effectively while ensuring high performance. Notably, our pipeline successfully identifies a substantial number of cancer-specific genes using a reduced feature set of just 500, in contrast to using the full dataset comprising 19,238 features. By employing an ensemble approach that combines three top-performing classifiers, a classification accuracy of 96.61% was achieved. Furthermore, we leverage Explainable AI to elucidate the biological significance of the identified cancer-specific genes, employing Differential Gene Expression (DGE) analysis.

Fairness in artificial intelligence and machine learning (AI/ML) models is becoming critically important, especially as decisions made by these systems impact diverse groups. In education, a vital sector for all countries, the widespread application of AI/ML systems raises specific concerns regarding fairness. Current research predominantly focuses on fairness for individual sensitive features, which limits the comprehensiveness of fairness assessments. This paper introduces FAIREDU, a novel and effective method designed to improve fairness across multiple sensitive features. Through extensive experiments, we evaluate FAIREDU effectiveness in enhancing fairness without compromising model performance. The results demonstrate that FAIREDU addresses intersectionality across features such as gender, race, age, and other sensitive features, outperforming state-of-the-art methods with minimal effect on model accuracy. The paper also explores potential future research directions to enhance further the method robustness and applicability to various machine-learning models and datasets.