Many machine learning problems, including similarity learning, ranking, and clustering, rely on empirical pairwise loss functions whose quadratic computational cost quickly becomes prohibitive at scale. We demonstrate how a frugal approach that retains only a fraction of the available information on pairs can achieve estimation or optimization performance comparable to that obtained by using all pairs, by leveraging survey sampling techniques. A central finding, supported by both theory and experiments, is that such sampling plans must target pairs directly rather than individual observations. In particular, for pairwise losses between high-dimensional vectors such as embeddings in vision or graph learning, assigning higher inclusion probabilities to informative pairs using suitable auxiliary information yields performance close to full pairwise evaluation, providing a principled and theoretically grounded trade-off between accuracy and computational cost.
Gender equality has been a pivotal focus in development agendas globally since the 1980s. Despite significant progress having been made thus far, patriarchal backlash remains a critical challenge in many regions across the globe, including Kenya. This study conducted in Kenya by Advocates for Social Change-Kenya (ADSOCK), explores the multifaceted dimensions of patriarchal backlash and resistance to gender equality initiatives, identifying key areas of concern and suggesting strategic interventions through which resistance to gender equality can be mitigated. The research has four main objectives: a) to understand the nature and extent of patriarchal backlash against gender equality in Kenya; b) to identify the socio-cultural, economic, and political factors contributing to patriarchal resistance and backlash; c) to assess the impact of patriarchal resistance on gender equality initiatives within communities and within institutions; and d) to recommend actionable solutions through which patriarchal backlash can be challenged and gender equality can be promoted.
This research briefing identifies how frontline workers delivering health, education, and social protection services in fragile and conflict-affected settings (FCAS) can be better supported. It synthesises evidence from six major research programmes that undertook studies across Africa, Asia, and the Middle East. The briefing’s core message is that resilient services depend on resilient frontline workforces, and that support for these workers during fragility and conflict must go beyond short-term incentives or fragmented projects. This review’s conclusion is that policymakers, donors, and practitioners need to take a more systemic, long-term and gender-transformative approach to supporting frontline workers if they want service delivery systems in FCAS to endure and strengthen more ambitious accountability and stability goals. The briefing also shows that evidence remains uneven and quite limited. There is emerging strong qualitative insight into recurrent problems and promising practices, but there is much more limited systematic evaluation of which support interventions for frontline workers are the most effective, affordable, and sustainable over time in different places and settings.
Human attention is the gateway to conscious perception, memory and decision-making. However, its role in modern transformer models remains largely unexplored. As these systems increasingly influence what people see, prefer and buy, the question arises as to whether they encode principles of human interest or merely exploit large-scale correlations. Addressing this issue is crucial for understanding cognition and ensuring the responsible use of AI in communication and marketing. In order to address this issue, the concept of visual interest was examined within the multimodal vision-language-model Qwen3-VL-8B, using a pre-defined Common Interestingness (CI) score derived from large-scale human engagement data on the photo-sharing platform Flickr. Here, we analyzed internal representations across vision and language components using methods from the neurosciences. Our analyses revealed that CI information is linearly decodable from final-layer embeddings, indicating that it is aligned with human-derived measures of visual interestingness. Dimensionality reduction and Generalized Discrimination Value (GDV) analyses demonstrate that CI-related hidden representations emerge in intermediate vision transformer layers and becomes progressively more distinguishable across language model layers. Concept vectors derived using geometric, probe, and Sparse Auto-Encoder based methods converge in higher layers, as confirmed by representational similarity analysis. This indicates a robust and structured encoding of visual interestingness without explicit supervision. Future work will seek to identify shared computational principles linking human brain dynamics and transformer architectures, with the ultimate goal of uncovering the organizing mechanisms that give rise to attention and interest in both biological and artificial systems.
