Evaluating personalized recommendations remains a central challenge, especially in long-form audio domains like podcasts, where traditional offline metrics suffer from exposure bias and online methods such as A/B testing are costly and operationally constrained. In this paper, we propose a novel framework that leverages Large Language Models (LLMs) as offline judges to assess the quality of podcast recommendations in a scalable and interpretable manner. Our two-stage profile-aware approach first constructs natural-language user profiles distilled from 90 days of listening history. These profiles summarize both topical interests and behavioral patterns, serving as compact, interpretable representations of user preferences. Rather than prompting the LLM with raw data, we use these profiles to provide high-level, semantically rich context-enabling the LLM to reason more effectively about alignment between a user's interests and recommended episodes. This reduces input complexity and improves interpretability. The LLM is then prompted to deliver fine-grained pointwise and pairwise judgments based on the profile-episode match. In a controlled study with 47 participants, our profile-aware judge matched human judgments with high fidelity and outperformed or matched a variant using raw listening histories. The framework enables efficient, profile-aware evaluation for iterative testing and model selection in recommender systems.

Existing video recommender systems rely primarily on user-defined metadata or on low-level visual and acoustic signals extracted by specialised encoders. These low-level features describe what appears on the screen but miss deeper semantics such as intent, humour, and world knowledge that make clips resonate with viewers. For example, is a 30-second clip simply a singer on a rooftop, or an ironic parody filmed amid the fairy chimneys of Cappadocia, Turkey? Such distinctions are critical to personalised recommendations yet remain invisible to traditional encoding pipelines. In this paper, we introduce a simple, recommendation system-agnostic zero-finetuning framework that injects high-level semantics into the recommendation pipeline by prompting an off-the-shelf Multimodal Large Language Model (MLLM) to summarise each clip into a rich natural-language description (e.g. "a superhero parody with slapstick fights and orchestral stabs"), bridging the gap between raw content and user intent. We use MLLM output with a state-of-the-art text encoder and feed it into standard collaborative, content-based, and generative recommenders. On the MicroLens-100K dataset, which emulates user interactions with TikTok-style videos, our framework consistently surpasses conventional video, audio, and metadata features in five representative models. Our findings highlight the promise of leveraging MLLMs as on-the-fly knowledge extractors to build more intent-aware video recommenders.

Generative retrieval for search and recommendation is a promising paradigm for retrieving items, offering an alternative to traditional methods that depend on external indexes and nearest-neighbor searches. Instead, generative models directly associate inputs with item IDs. Given the breakthroughs of Large Language Models (LLMs), these generative systems can play a crucial role in centralizing a variety of Information Retrieval (IR) tasks in a single model that performs tasks such as query understanding, retrieval, recommendation, explanation, re-ranking, and response generation. Despite the growing interest in such a unified generative approach for IR systems, the advantages of using a single, multi-task model over multiple specialized models are not well established in the literature. This paper investigates whether and when such a unified approach can outperform task-specific models in the IR tasks of search and recommendation, broadly co-existing in multiple industrial online platforms, such as Spotify, YouTube, and Netflix. Previous work shows that (1) the latent representations of items learned by generative recommenders are biased towards popularity, and (2) content-based and collaborative-filtering-based information can improve an item’s representations. Motivated by this, our study is guided by two hypotheses: [H1] the joint training regularizes the estimation of each item’s popularity, and [H2] the joint training regularizes the item’s latent representations, where search captures content-based aspects of an item and recommendation captures collaborative-filtering aspects. Our extensive experiments with both simulated and real-world data support both [H1] and [H2] as key contributors to the effectiveness improvements observed in the unified search and recommendation generative models over the single-task approaches.

Abstract Recent work on Vision Transformers (VTs) showed that introducing a local inductive bias in the VT architecture helps reducing the number of samples necessary for training. However, the architecture modifications lead to a loss of generality of the Transformer backbone, partially contradicting the push towards the development of uniform architectures, shared, e.g., by both the Computer Vision and the Natural Language Processing areas. In this work, we propose a different and complementary direction, in which a local bias is introduced using an auxiliary self-supervised task, performed jointly with standard supervised training. Specifically, we exploit the observation that the attention maps of VTs, when trained with self-supervision, can contain a semantic segmentation structure which does not spontaneously emerge when training is supervised. Thus, we explicitly encourage the emergence of this spatial clustering as a form of training regularization. In more detail, we exploit the assumption that, in a given image, objects usually correspond to few connected regions, and we propose a spatial formulation of the information entropy to quantify this object-based inductive bias. By minimizing the proposed spatial entropy, we include an additional self-supervised signal during training. Using extensive experiments, we show that the proposed regularization leads to equivalent or better results than other VT proposals which include a local bias by changing the basic Transformer architecture, and it can drastically boost the VT final accuracy when using small-medium training sets. The code is available at https://github.com/helia95/SAR .

Recently, there has been an increasing interest in image editing methods that employ pre-trained unconditional image generators (e.g., StyleGAN). However, applying these methods to translate images to multiple visual domains remains challenging. Existing works do not often preserve the domain-invariant part of the image (e.g., the identity in human face translations), or they do not usually handle multiple domains or allow for multi-modal translations. This work proposes an implicit style function (ISF) to straightforwardly achieve multi-modal and multi-domain image-to-image translation from pre-trained unconditional generators. The ISF manipulates the semantics of a latent code to ensure that the image generated from the manipulated code lies in the desired visual domain. Our human faces and animal image manipulations show significantly improved results over the baselines. Our model enables cost-effective multi-modal unsupervised image-to-image translations at high resolution using pre-trained unconditional GANs. The code and data are available at:https://github.com/yhlleo/stylegan-mmuit.

Image-to-Image (I2I) multi-domain translation models are usually evaluated also using the quality of their semantic interpolation results. However, state-of-the-art models frequently show abrupt changes in the image appearance during interpolation, and usually perform poorly in interpolations across domains. In this paper, we propose a new training protocol based on three specific losses which help a translation network to learn a smooth and disentangled latent style space in which: 1) Both intra- and inter-domain interpolations correspond to gradual changes in the generated images and 2) The content of the source image is better preserved during the translation. Moreover, we propose a novel evaluation metric to properly measure the smoothness of latent style space of I2I translation models. The proposed method can be plugged in existing translation approaches, and our extensive experiments on different datasets show that it can significantly boost the quality of the generated images and the graduality of the interpolations.

This paper introduces Click to Move (C2M), a novel framework for video generation where the user can control the motion of the synthesized video through mouse clicks specifying simple object trajectories of the key objects in the scene. Our model receives as input an initial frame, its corresponding segmentation map and the sparse motion vectors encoding the input provided by the user. It outputs a plausible video sequence starting from the given frame and with a motion that is consistent with user input. Notably, our proposed deep architecture incorporates a Graph Convolution Network (GCN) modelling the movements of all the objects in the scene in a holistic manner and effectively combining the sparse user motion information and image features. Experimental results show that C2M outperforms existing methods on two publicly available datasets, thus demonstrating the effectiveness of our GCN framework at modelling object interactions. The source code is publicly available at https://github.com/PierfrancescoArdino/C2M.

Visual Transformers (VTs) are emerging as an architectural paradigm alternative to Convolutional networks (CNNs). Differently from CNNs, VTs can capture global relations between image elements and they potentially have a larger representation capacity. However, the lack of the typical convolutional inductive bias makes these models more data hungry than common CNNs. In fact, some local properties of the visual domain which are embedded in the CNN architectural design, in VTs should be learned from samples. In this paper, we empirically analyse different VTs, comparing their robustness in a small training set regime, and we show that, despite having a comparable accuracy when trained on ImageNet, their performance on smaller datasets can be largely different. Moreover, we propose an auxiliary self-supervised task which can extract additional information from images with only a negligible computational overhead. This task encourages the VTs to learn spatial relations within an image and makes the VT training much more robust when training data is scarce. Our task is used jointly with the standard (supervised) training and it does not depend on specific architectural choices, thus it can be easily plugged in the existing VTs. Using an extensive evaluation with different VTs and datasets, we show that our method can improve (sometimes dramatically) the final accuracy of the VTs. Our code is available at: https://github.com/yhlleo/VTs-Drloc.

As "Big Data" has become pervasive, an increasing amount of research has connected the dots between human behaviour in the offline and online worlds. Consequently, researchers have exploited these new findings to create models that better predict different aspects of human life and recommend future behaviour. To date, however, we do not yet fully understand the similarities and differences of human behaviour in these virtual and physical worlds. Here, we analyse and discuss the mobility and application usage of 400,000 individuals over eight months. We find an astonishing similarity between people's mobility in the physical space and how they move from app to app in smartphones. Our data shows that individuals use and visit a finite number of apps and places, but they keep exploring over time. In particular, two distinct profiles of individuals emerge: those that keep changing places and services, and those that are stable over time, named as "explorers" and "keepers". We see these findings as crucial to enrich a discussion for the potentials and the challenges of building human-centric AI systems, which might leverage recent results in Computational Social Science.

Manipulating images of complex scenes to reconstruct, insert and/or remove specific object instances is a challenging task. Complex scenes contain multiple semantics and objects, which are frequently cluttered or ambiguous, thus hampering the performance of inpainting models. Conventional techniques often rely on structural information such as object contours in multi-stage approaches that generate unreliable results and boundaries. In this work, we propose a novel deep learning model to alter a complex urban scene by removing a user-specified portion of the image and coherently inserting a new object (e.g. car or pedestrian) in that scene. Inspired by recent works on image inpainting, our proposed method leverages the semantic segmentation to model the content and structure of the image, and learn the best shape and location of the object to insert. To generate reliable results, we design a new decoder block that combines the semantic segmentation and generation task to guide better the generation of new objects and scenes, which have to be semantically consistent with the image. Our experiments, conducted on two large-scale datasets of urban scenes (Cityscapes and Indian Driving), show that our proposed approach successfully address the problem of semantically-guided inpainting of complex urban scene.

Image to image translation aims to learn a mapping that transforms an image from one visual domain to another. Recent works assume that images descriptors can be disentangled into a domain-invariant content representation and a domain-specific style representation. Thus, translation models seek to preserve the content of source images while changing the style to a target visual domain. However, synthesizing new images is extremely challenging especially in multi-domain translations, as the network has to compose content and style to generate reliable and diverse images in multiple domains. In this paper we propose the use of an image retrieval system to assist the image-to-image translation task. First, we train an image-to-image translation model to map images to multiple domains. Then, we train an image retrieval model using real and generated images to find images similar to a query one in content but in a different domain. Finally, we exploit the image retrieval system to fine-tune the image-to-image translation model and generate higher quality images. Our experiments show the effectiveness of the proposed solution and highlight the contribution of the retrieval network, which can benefit from additional unlabeled data and help image-to-image translation models in the presence of scarce data.

Manipulating visual attributes of images through human-written text is a very challenging task. On the one hand, models have to learn the manipulation without the ground truth of the desired output. On the other hand, models have to deal with the inherent ambiguity of natural language. Previous research usually requires either the user to describe all the characteristics of the desired image or to use richly-annotated image captioning datasets. In this work, we propose a novel unsupervised approach, based on image-to-image translation, that alters the attributes of a given image through a command-like sentence such as "change the hair color to black". Contrarily to state-of-the-art approaches, our model does not require a human-annotated dataset nor a textual description of all the attributes of the desired image, but only those that have to be modified. Our proposed model disentangles the image content from the visual attributes, and it learns to modify the latter using the textual description, before generating a new image from the content and the modified attribute representation. Because text might be inherently ambiguous (blond hair may refer to different shadows of blond, e.g. golden, icy, sandy), our method generates multiple stochastic versions of the same translation. Experiments show that the proposed model achieves promising performances on two large-scale public datasets: CelebA and CUB. We believe our approach will pave the way to new avenues of research combining textual and speech commands with visual attributes.

Recent works have shown Generative Adversarial Networks (GANs) to be particularly effective in image-to-image translations. However, in tasks such as body pose and hand gesture translation, existing methods usually require precise annotations, e.g. key-points or skeletons, which are time-consuming to draw. In this work, we propose a novel GAN architecture that decouples the required annotations into a category label - that specifies the gesture type - and a simple-to-draw category-independent conditional map - that expresses the location, rotation and size of the hand gesture. Our architecture synthesizes the target gesture while preserving the background context, thus effectively dealing with gesture translation in the wild. To this aim, we use an attention module and a rolling guidance approach, which loops the generated images back into the network and produces higher quality images compared to competing works. Thus, our GAN learns to generate new images from simple annotations without requiring key-points or skeleton labels. Results on two public datasets show that our method outperforms state of the art approaches both quantitatively and qualitatively. To the best of our knowledge, no work so far has addressed the gesture-to-gesture translation in the wild by requiring user-friendly annotations.

Housing costs have a significant impact on individuals, families, businesses, and governments. Recently, online companies such as Zillow have developed proprietary systems that provide automated estimates of housing prices without the immediate need of professional appraisers. Yet, our understanding of what drives the value of houses is very limited. In this paper, we use multiple sources of data to entangle the economic contribution of the neighborhood's characteristics such as walkability and security perception. We also develop and release a framework able to now-cast housing prices from Open data, without the need for historical transactions. Experiments involving 70,000 houses in 8 Italian cities highlight that the neighborhood's vitality and walkability seem to drive more than 20% of the housing value. Moreover, the use of this information improves the nowcast by 60%. Hence, the use of property's surroundings' characteristics can be an invaluable resource to appraise the economic and social value of houses after neighborhood changes and, potentially, anticipate gentrification.

Policy makers, urban planners, architects, sociologists, and economists are interested in creating urban areas that are both lively and safe. But are the safety and liveliness of neighborhoods independent characteristics? Or are they just two sides of the same coin? In a world where people avoid unsafe looking places, neighborhoods that look unsafe will be less lively, and will fail to harness the natural surveillance of human activity. But in a world where the preference for safe looking neighborhoods is small, the connection between the perception of safety and liveliness will be either weak or nonexistent. In this paper we explore the connection between the levels of activity and the perception of safety of neighborhoods in two major Italian cities by combining mobile phone data (as a proxy for activity or liveliness) with scores of perceived safety estimated using a Convolutional Neural Network trained on a dataset of Google Street View images scored using a crowdsourced visual perception survey. We find that: (i) safer looking neighborhoods are more active than what is expected from their population density, employee density, and distance to the city centre; and (ii) that the correlation between appearance of safety and activity is positive, strong, and significant, for females and people over 50, but negative for people under 30, suggesting that the behavioral impact of perception depends on the demographic of the population. Finally, we use occlusion techniques to identify the urban features that contribute to the appearance of safety, finding that greenery and street facing windows contribute to a positive appearance of safety (in agreement with Oscar Newman's defensible space theory). These results suggest that urban appearance modulates levels of human activity and, consequently, a neighborhood's rate of natural surveillance.

The Death and Life of Great American Cities was written in 1961 and is now one of the most influential book in city planning. In it, Jane Jacobs proposed four conditions that promote life in a city. However, these conditions have not been empirically tested until recently. This is mainly because it is hard to collect data about "city life". The city of Seoul recently collected pedestrian activity through surveys at an unprecedented scale, with an effort spanning more than a decade, allowing researchers to conduct the first study successfully testing Jacobs's conditions.