Martha Larson

Anna Pillar, Kyrill Poelmans, Martha Larson

Deep learning holds great promise for detecting discriminatory language in the public sphere. However, for the detection of illegal age discrimination in job advertisements, regex approaches are still strong performers. In this paper, we investigate job advertisements in the Netherlands. We present a qualitative analysis of the benefits of the 'old' approach based on regexes and investigate how neural embeddings could address its limitations.

Manel Slokom, Özlem Özgöbek, Martha Larson

This paper studies user attributes in light of current concerns in the recommender system community: diversity, coverage, calibration, and data minimization. In experiments with a conventional context-aware recommender system that leverages side information, we show that user attributes do not always improve recommendation. Then, we demonstrate that user attributes can negatively impact diversity and coverage. Finally, we investigate the amount of information about users that ``survives'' from the training data into the recommendation lists produced by the recommender. This information is a weak signal that could in the future be exploited for calibration or studied further as a privacy leak.

Manel Slokom, Peter-Paul de Wolf, Martha Larson

We investigate an attack on a machine learning model that predicts whether a person or household will relocate in the next two years, i.e., a propensity-to-move classifier. The attack assumes that the attacker can query the model to obtain predictions and that the marginal distribution of the data on which the model was trained is publicly available. The attack also assumes that the attacker has obtained the values of non-sensitive attributes for a certain number of target individuals. The objective of the attack is to infer the values of sensitive attributes for these target individuals. We explore how replacing the original data with synthetic data when training the model impacts how successfully the attacker can infer sensitive attributes.

Yana van de Sande, Gunes Açar, Thabo van Woudenberg et al.

We study LLM judgments of misinformation expressed with uncertainty. Our experiments study the response of three widely used LLMs (GPT-4o, LlaMA3, DeepSeek-v2) to misinformation propositions that have been verified false and then are transformed into uncertain statements according to an uncertainty typology. Our results show that after transformation, LLMs change their factchecking classification from false to not-false in 25% of the cases. Analysis reveals that the change cannot be explained by predictors to which humans are expected to be sensitive, i.e., modality, linguistic cues, or argumentation strategy. The exception is doxastic transformations, which use linguistic cue phrases such as "It is believed ...".To gain further insight, we prompt the LLM to make another judgment about the transformed misinformation statements that is not related to truth value. Specifically, we study LLM estimates of the frequency with which people make the uncertain statement. We find a small but significant correlation between judgment of fact and estimation of frequency.

Mingliang Liang, Martha Larson

Vision Language Models (VLMs) can be trained more efficiently if training sets can be reduced in size. Recent work has shown the benefits of masking text during VLM training using a variety of approaches: truncation, random masking, block masking and syntax masking. In this paper, we show that the best masking strategy changes over training epochs and that, given sufficient training epochs. We analyze existing text masking approaches including syntax masking, which is currently the state of the art, and identify the word frequency distribution as important in determining their success. Experiments on a large range of data sets demonstrate that syntax masking is outperformed by other approaches, given sufficient epochs, and that our proposed frequency-based approach, called Contrastive Language-Image Pre-training with Word Frequency Masking (CLIPF) has numerous advantages. The benefits are particularly evident as the number of input tokens decreases.

Zhengyu Zhao, Martha Larson

As deep learning approaches to scene recognition emerge, they have continued to leverage discriminative regions at multiple scales, building on practices established by conventional image classification research. However, approaches remain largely generic, and do not carefully consider the special properties of scenes. In this paper, inspired by the intuitive differences between scenes and objects, we propose Adi-Red, an adaptive approach to discriminative region discovery for scene recognition. Adi-Red uses a CNN classifier, which was pre-trained using only image-level scene labels, to discover discriminative image regions directly. These regions are then used as a source of features to perform scene recognition. The use of the CNN classifier makes it possible to adapt the number of discriminative regions per image using a simple, yet elegant, threshold, at relatively low computational cost. Experimental results on the scene recognition benchmark dataset SUN397 demonstrate the ability of Adi-Red to outperform the state of the art. Additional experimental analysis on the Places dataset reveals the advantages of Adi-Red, and highlight how they are specific to scenes. We attribute the effectiveness of Adi-Red to the ability of adaptive region discovery to avoid introducing noise, while also not missing out on important information.

Jeroen B. P. Vuurens, Martha Larson, Arjen P. de Vries

Recommender systems leverage both content and user interactions to generate recommendations that fit users' preferences. The recent surge of interest in deep learning presents new opportunities for exploiting these two sources of information. To recommend items we propose to first learn a user-independent high-dimensional semantic space in which items are positioned according to their substitutability, and then learn a user-specific transformation function to transform this space into a ranking according to the user's past preferences. An advantage of the proposed architecture is that it can be used to effectively recommend items using either content that describes the items or user-item ratings. We show that this approach significantly outperforms state-of-the-art recommender systems on the MovieLens 1M dataset.

Mark Melenhorst, María Menéndez Blanco, Martha Larson

Research on mid-level image representations has conventionally concentrated relatively obvious attributes and overlooked non-obvious attributes, i.e., characteristics that are not readily observable when images are viewed independently of their context or function. Non-obvious attributes are not necessarily easily nameable, but nonetheless they play a systematic role in people`s interpretation of images. Clusters of related non-obvious attributes, called interpretation dimensions, emerge when people are asked to compare images, and provide important insight on aspects of social images that are considered relevant. In contrast to aesthetic or affective approaches to image analysis, non-obvious attributes are not related to the personal perspective of the viewer. Instead, they encode a conventional understanding of the world, which is tacit, rather than explicitly expressed. This paper introduces a procedure for discovering non-obvious attributes using crowdsourcing. We discuss this procedure using a concrete example of a crowdsourcing task on Amazon Mechanical Turk carried out in the domain of fashion. An analysis comparing discovered non-obvious attributes with user tags demonstrated the added value delivered by our procedure.

Yue Shi, Alexandros Karatzoglou, Linas Baltrunas et al.

Recommender systems are frequently used in domains in which users express their preferences in the form of graded judgments, such as ratings. If accurate top-N recommendation lists are to be produced for such graded relevance domains, it is critical to generate a ranked list of recommended items directly rather than predicting ratings. Current techniques choose one of two sub-optimal approaches: either they optimize for a binary metric such as Average Precision, which discards information on relevance grades, or they optimize for Normalized Discounted Cumulative Gain (NDCG), which ignores the dependence of an item's contribution on the relevance of more highly ranked items. In this paper, we address the shortcomings of existing approaches by proposing the Graded Average Precision factor model (GAPfm), a latent factor model that is particularly suited to the problem of top-N recommendation in domains with graded relevance data. The model optimizes for Graded Average Precision, a metric that has been proposed recently for assessing the quality of ranked results list for graded relevance. GAPfm learns a latent factor model by directly optimizing a smoothed approximation of GAP. GAPfm's advantages are twofold: it maintains full information about graded relevance and also addresses the limitations of models that optimize NDCG. Experimental results show that GAPfm achieves substantial improvements on the top-N recommendation task, compared to several state-of-the-art approaches. In order to ensure that GAPfm is able to scale to very large data sets, we propose a fast learning algorithm that uses an adaptive item selection strategy. A final experiment shows that GAPfm is useful not only for generating recommendation lists, but also for ranking a given list of rated items.