In this paper, we present TubeLens, a system designed to support YouTube users in reflecting on how recommendation algorithms perceive and represent their interests. TubeLens invites users to engage with their algorithmic selves through self-portraits accompanied by dispositional keywords and explanations, creating space to consider how algorithmic experiences might be interpreted and potentially reshaped over time. Rather than positioning users as passive recipients of recommendations, TubeLens foregrounds users’ agency in questioning and making sense of algorithmic influence on their media consumption. We conducted an exploratory user study with 22 participants to examine users’ experiences with TubeLens. Our findings suggest that algorithmic self-portraits can surface gaps between perceived and algorithmic selves, supporting self-awareness and agentic awareness, while also revealing tensions around privacy and social comparison. This work offers initial insights into how interactive representations of algorithmic profiles can support reflective engagement with algorithmic systems and inform the design of future identity-oriented interfaces.

Creative face stylization aims to render portraits in diverse visual idioms such as cartoons, sketches, and paintings while retaining recognizable identity. However, current identity encoders, which are typically trained and calibrated on natural photographs, exhibit severe brittleness under stylization. They often mistake changes in texture or color palette for identity drift or fail to detect geometric exaggerations. This reveals the lack of a style-agnostic framework to evaluate and supervise identity consistency across varying styles and strengths. To address this gap, we introduce StyleID, a human perception-aware dataset and evaluation framework for facial identity under stylization. StyleID comprises two datasets: (i) StyleBench-H, a benchmark that captures human same-different verification judgments across diffusion- and flow-matching-based stylization at multiple style strengths, and (ii) StyleBench-S, a supervision set derived from psychometric recognition-strength curves obtained through controlled two-alternative forced-choice (2AFC) experiments. Leveraging StyleBench-S, we fine-tune existing semantic encoders to align their similarity orderings with human perception across styles and strengths. Experiments demonstrate that our calibrated models yield significantly higher correlation with human judgments and enhanced robustness for out-of-domain, artist drawn portraits. All of our datasets, code, and pretrained models will be publicly available.

In the rapidly evolving field of deep learning, traditional methodologies for designing models predominantly rely on code-based frameworks. While these approaches provide flexibility, they create a significant barrier to entry for non-experts and obscure the immediate impact of architectural decisions on model performance. In response to this challenge, recent no-code approaches have been developed with the aim of enabling easy model development through graphical interfaces. However, both traditional and no-code methodologies share a common limitation that the inability to predict model outcomes or identify issues without executing the model. To address this limitation, we introduce an intuitive visual feedback-based no-code approach to visualize and analyze deep learning models during the design phase. This approach utilizes dataflow-based visual programming with dynamic visual encoding of model architecture. A user study was conducted with deep learning developers to demonstrate the effectiveness of our approach in enhancing the model design process, improving model understanding, and facilitating a more intuitive development experience. The findings of this study suggest that real-time architectural visualization significantly contributes to more efficient model development and a deeper understanding of model behaviors.

The offline pickup and delivery problem with time windows (PDPTW) is a classical combinatorial optimization problem in the transportation community, which has proven to be very challenging computationally. Due to the complexity of the problem, practical problem instances can be solved only via heuristics, which trade-off solution quality for computational tractability. Among the various heuristics, a common strategy is problem decomposition, that is, the reduction of a large-scale problem into a collection of smaller sub-problems, with spatial and temporal decompositions being two natural approaches. While spatial decomposition has been successful in certain settings, effective temporal decomposition has been challenging due to the difficulty of stitching together the sub-problem solutions across the decomposition boundaries. In this work, we introduce a novel temporal decomposition scheme for solving a class of PDPTWs that have narrow time windows, for which it is able to provide both fast and high-quality solutions. We utilize techniques that have been popularized recently in the context of online dial-a-ride problems along with the general idea of rolling horizon optimization. To the best of our knowledge, this is the first attempt to solve offline PDPTWs using such an approach. To show the performance and scalability of our framework, we use the optimization of paratransit services as a motivating example. Due to the lack of benchmark solvers similar to ours (i.e., temporal decomposition with an online solver), we compare our results with an offline heuristic algorithm using Google OR-Tools. In smaller problem instances (with an average of 129 requests per instance), the baseline approach is as competitive as our framework. However, in larger problem instances (approximately 2,500 requests per instance), our framework is more scalable and can provide good solutions to problem instances of varying degrees of difficulty, while the baseline algorithm often fails to find a feasible solution within comparable compute times.