Abstract

Through an initiative spearheaded by its most populous community--the Dawoodi Bohras--Mumbai's historic Bhendi Bazaar locality is being redeveloped by the Saifee Burhani Upliftment Trust (SBUT). What happens to a neighborhood and its inhabitants as they experience such a colossal de/reassembling? How are old homes let go of and new ones fashioned? What does one do as they "wait" for the state of transition to end? What role does/can faith play in shaping both ideas of aspiring for progress and holding on to history? I entered the field in the summer of 2022 with these questions to understand the ways in which the Bohras of Bhendi Bazaar are experiencing and articulating the ongoing material transformation of their lived environment and how it fits within as well as embodies a larger discourse of "upliftment" that the community universally ascribes to. At the same time, the aim was also to examine how the Bohras engage with and navigate Bhendi Bazaar on an everyday basis through the conceptual lens of translocality, which enables notions of belonging that are simultaneously mobile and grounded. Based on ethnographic fieldwork conducted between 2022 to 2024, this article argues that the emergent picture is a complex one of a neighborhood long marked by histories of both conflict and cooperation as well as a community straddling discourses of inclusion and exclusivity. In doing so, it complicates and nuances established ideas of place-making in urban spaces that have long histories of diverse communities inhabiting them with enmeshed lives and livelihoods as well as of specific ones--such as the Bohras--asserting a distinctive identity and heritage through the built environment while simultaneously claiming the transition to be a seamless one.

Abstract

Corrosion creates a significant degradation mechanism in reinforced concrete (RC) structures, which would require a high cost of maintenance and repair in affected buildings. The right methods for structural retrofitting are, therefore, quite pivotal. Modification or rectification for any structure is normally categorized under three major categories: (i) Structural rehabilitation, (ii) Structural refurbishment, and (iii) Structural retrofit. All these activities have a thin but predominant line of differentiation. In contrast, the first two activities relate to enhancing the functional life of a structure that has been damaged under an adverse environment or condition. On the other hand, structural retrofit is an activity where the structure is updated in its basic parameters of sharing the load and supporting the superstructure. It improves the safety and durability of the structure. Further, when corrosion hits the structure, it results in a reduction in many important parameters to the extent that they go below their acceptable limits under national and international standards. It is indeed a perfect blend of structural dynamics or design engineering and functionality of the structure in the right proportion when retrofit action for a corrosion-hit structure is conceptualized and crafted. This paper aims to present and discuss various structural retrofitting methods for RC members affected by corrosion, supported by relevant real-world case studies.

Abstract

Medical image segmentation is often ambiguous: unclear bound-
aries, image artifacts, and genuine inter-observer disagreement

mean that no single mask is always the only valid target. Determin-
istic segmentation models collapse this ambiguity into one output,

discarding clinically meaningful variability. We present Ambigu-
ousTextDiff, a text-guided diffusion framework that generates a

diverse set of plausible segmentation masks per image, conditioned
jointly on the image and natural-language descriptors assembled

from structured clinical metadata (e.g., nodule texture and spic-
ulation, lesion site, and pathology metadata). For each image, a

single case-level text description is held fixed while repeated sto-
chastic diffusion sampling generates multiple plausible masks, so

the ensemble can represent the range of segmentations different
radiologists may provide for the same case. The model applies

diffusion only to the mask, fuses the image via channel-wise con-
catenation, and fuses text via cross-attention injected at multiple

U-Net resolutions; a latent KL term, adapted from the Probabilistic

U-Net, regularizes the model toward the distribution of annota-
tor masks rather than a single sampled target. We evaluate on the

complete, unfiltered LIDC-IDRI lung-CT test split (3,072 images)
and on the IMA++ skin-dermoscopy benchmark (2,394 images),

reporting both ambiguity-aware metrics (Generalized Energy Dis-
tance, Collective Insight) and standard overlap metrics. On full

LIDC-IDRI, AmbiguousTextDiff improves GED from 0.306 to 0.152
and CI from 0.470 to 0.835 over the CIMD diffusion baseline; on
IMA++, it improves GED from 0.139 to 0.072 and CI from 0.796

to 0.885 over the strongest baseline. The framework targets set-
tings where structured descriptors are already available before or

alongside pixel-level delineation (e.g., report-conditioned contour-
ing, retrospective cohort curation, or multidisciplinary planning),

and we report controlled ablations, a same-prompt TextDiff base-
line, a descriptor ablation, paired-bootstrap confidence intervals, a

regularization-weight sweep, and a DDPM/DDIM inference-cost
comparison.

Abstract

Visual dataset auditors rank image-label pairs for human review
using signals such as classifier confidence, training dynamics, rep-
resentation geometry, neighborhood agreement, or image-text com-
patibility. However, these signals can confuse annotation error with
visual atypicality: a correctly labeled long-tail or domain-shifted
example may appear suspicious simply because it is rare. We call
this the rare-or-wrong problem.
We introduce BiSafeAudit, a finite selector for deciding when an
alternative auditor should replace a calibrated baseline anchor. Us-
ing a verified calibration subset, BiSafeAudit builds a fixed library
of rank-normalized auditing scores and pairwise rank ensembles. A
candidate may replace the anchor only if pooled AUPRC and worst-
group error recall remain within fixed anchor-relative tolerances,
with the recall requirement also checked across leave-one-fold-
out calibration subsets. Among qualified candidates, BiSafeAudit
minimizes worst-group false-positive rate; otherwise, it retains the
anchor.
Across four long-tailed or shifted benchmarks and five seeds
(20 held-out evaluations), ranking-only and unconstrained worst-
group-FPR selection each produce seven anchor-relative down-
side events, pooled qualification reduces the count to four, and
the complete selector reduces it to two. BiSafeAudit still selects
a non-anchor auditor in seven runs, showing that conservative
qualification can admit complementary evidence while avoiding
most observed harmful replacements.

Abstract

Motion planning in robotics requires not only computing collision-free paths but also certifying infeasibility when no such path exists. Complete methods are limited to low-dimensional spaces, while sampling-based planners scale efficiently but cannot provide finite-time infeasibility certificates, leaving this problem largely unresolved in high-dimensional spaces. In this letter, we present a geometry-driven framework for certifying infeasibility through an explicit resolution-dependent analysis of configuration space topology. Leveraging signed distance field representations, the proposed method traces separating manifolds induced by obstacle boundaries directly in configuration space, enabling both detection of infeasibility and identification of the specific geometric cause. To address computational challenges, we develop a parallel frontier-expansion algorithm that exploits GPU acceleration for efficient simplicial reconstruction in high-dimensional spaces. We validate the approach on 4-DOF and 5-DOF robot scenarios, certifying infeasibility within seconds for 4-DOF cases and under four minutes for 5-DOF cases. We further discuss avenues for improving scalability to higher-dimensional spaces.

Abstract

On-device intelligence for embodied agents must operate under highly dynamic resource constraints caused by shifting concurrent workloads and thermal throttling. Static deployment strategies either risk severe memory and latency violations or conservatively sacrifice task accuracy. We present ElasticInfer, an adaptive framework that decouples pre-execution resource admission from online model selection. By combining a Nano-first feasibility filter with explicit delivery accounting and cooperative allocation leases, ElasticInfer prevents catastrophic out-of-memory (OOM) and deadline failures, while dynamically optimizing perception quality across a portfolio of detectors. Evaluated on 5,000 COCO validation images under simulated resource dynamics, ElasticInfer achieves 0.406 mAP at a mean latency of 12.3 ms, utilizing contextual Gaussian Thompson sampling to provide label-free exploration of the remaining accuracy-cost trade-off. In a physical GPU evaluation featuring concurrent compute, managed allocation, and warmed input shapes, the guarded admission protocol successfully delivers 100% of its 180 queries within a strict 50 ms deadline (averaging 21.31 ms) with zero observed system violations. These results demonstrate that explicit resource admission and selective context acquisition are the primary drivers of safe edge inference, allowing embodied agents to gracefully defer queries rather than suffer unregulated system crashes.

Abstract

While acting early on conversational cues makes a robot feel responsive, premature movement can introduce physical hazards. The root challenge is that recognizing a human's intent and authorizing robotic motion are separate decisions. We solve this with the Risk-Aware Incremental Action Policy (RIAP), a framework that decouples these decisions by weighing real-time intent confidence against action- and state-dependent safety costs. RIAP continually selects the optimal behavior from three choices: ACT, WAIT, or DEFER. Through 8,160 simulation replays, RIAP cut unsafe commitments from 27.5% to 1.6%, anticipated user commands by an average of 1,022 ms, and corrected for 60.6% of mid-utterance human revisions--a trade-off profile that existing decision mechanisms could not reproduce. A 30-participant study further confirms that RIAP delivers a significantly superior perceived user experience across all measured subjective dimensions.

Abstract

Dawoodi Bohra er en ismailitisk minoritet, som i dag har sit centrum i Mumbai og Indien, men som historisk trækker tråde tilbage til det fatimidiske Cairo, og desuden består af en række diaspora-grupper forskellige steder i verden, især omkring det Indiske Ocean. Denne artikel handler om bohraerne i det vestligeIndiske Ocean, deres tætte transregionale forbindelser, og særligt hvordan disse
relaterer sig til diverse infrastrukturer, som binder bohraer sammen lokalt og transregionalt - moskebyggerier og -restaureringer, boligblokke, skoler, universiteter, hospitaler med mere. Således bygger artiklen på nyere diskussioner af infrastruktur som humanistisk forskningsfelt, samt på begrebet 'tæt transregionalisme' (thick transregionalism) foreslået af Indisk Ocean-forskeren Engseng Ho. Artiklen bygger på det kollektive og interdisciplinære forskningsprojekt Constructing the Ocean: Indian Ocean Infrastructures and Thick Transregionalism of Dawoodi Bohras (CO-OC) og præsenterer dets teoretiske og metodiske rammer samt de forskellige felter, vi særligt har undersøgt: Mumbai (Isha Dubey), Cairo (Mattias Gori Olesen), Tanzania (Cecil Schou Pallesen), Dubai (Thomas Fibiger), mens Uwe Skoda i sit syntetiserende delprojekt især har fokuseret på visualitet og arkitektur på tværs af disse forskellige lokationer. Målet med projektet har dog i høj grad været at sammentænke de fire lokationer i en integreret, tværfaglig undersøgelse, der kombinerer bådehistoriske og etnografiske metoder (især arkiv- og feltarbejde), og hvor også gensidige
besøg på alle feltsteder har tjent som et metodisk værktøj til at styrke forståelsen af de mange tværregionale forbindelser
og perspektiver.

Abstract

The year 2020-21(from March 17, 2020, to March 26 2021) marked the birth centenary of Sheikh Mujibur Rahman, the founding leader of Bangladesh and its first Prime Minister. It was declared as Mujib Borsho or the 'Year of Mujib' at a grand opening ceremony as the countdown was kickstarted by Prime Minister and the elder daughter of Mujibur Rahman, Sheikh Haseena at the historic Parade Square where he had landed in the newly liberated Bangladesh on January 10, 1972, from Pakistan. In the speech delivered by her at this inaugural event, she talked at length about the collective past, present and future of Bangladesh and how its story was inextricably linked to the life and works of the founder of the nation. And while much of the address focused on the significance of celebrating the extraordinary life of Sheikh Mujibur Rahman, it was also steeped in an aura of lament for the fact that he was so brutally lost to the nation only four years after its liberation and what that loss has meant to its sense of self. And so, the idea of mourning is inherent and inbuilt into the idea of celebrating the 'year of Mujib'. Similarly, and relatedly, the running thread through much of the cultural memory of the Bangladeshi nation is the simultaneous celebration of its Liberation War, the secular ideals which informed it and the courage and resilience of the Bengali people who fought it as well as a deep sense of mourning for the violence and death that marked the birth of Bangladesh and in fact continued with the assassination of Sheikh Mujibur Rahman. The common approach to conceptualising trauma in historical interpretations of the past is to perceive it as a 'broken time' that needs to be accounted for so as to be able to learn and move forward from. This is especially the case with historical scholarship on the Holocaust that sees it as a break of civilization. On the other hand, the cultural memories of post-colonial nations such as India etch out a narrative of triumph and celebration of the struggle that led up to it. However, in Bangladesh the idea of mourning, pain and loss is central to remembering and commemorating its birth to the extent that the collective trauma of the war and its aftermath is a living, breathing entity that runs through much of the memorial landscape of the nation. Hence rather than embodying a so- called civilizational break in the onward march of history, trauma and the act of mourning this trauma is something that confers meaning to the narrative of the Bangladeshi nation and its sense of self. This paper looks at the ways in and the extent to which this sense of continuing collective trauma and mourning is deployed in curating the national cultural memory of Bangladesh and the responses it evokes in those who navigate it. This shall be done firstly by analysing the narrative choices adopted by four of the most prominent museums in the capital city of Dhaka - the National Museum of Bangladesh, the Museum of Independence and the Liberation War Museum - for constructing, showcasing and memorialising the nation's history. Secondly, the paper examines the nature of public engagement with this consciously curated cultural memory at these four museums and also across other popular sites of memory associated with the Liberation War in the city.

Abstract

This volume explores new avenues in the field of memory studies. Going beyond Western frames of reference, it shows how religion, region, caste, and class and their intersectionality and the fraught legacy of colonialism shape acts of both collective remembrance and forgetting in South Asia. The chapters in the volume call for a substantive rethinking of the conceptual and methodological frameworks in the field by posing critical questions such as, how do these layers play into and inform the processes of crafting and curating national histories and memories in South Asia? What are the silences that exist within it and how are they contested? What are the alternative modes of remembering, marking and accounting for "difficult pasts" beyond the confines of state-regulated memorial projects? And what "events" constitute dominant and rightful entry points for engaging with these themes and which remain ignored? In doing so they steer discussions on the politics of memory in the region in directions that offer opportunities for not only re-visiting the Partition of 1947 from previously unexplored perspectives but also for going beyond it as the central analytical lens for approaching questions of remembrance, forgetting and utterance in South Asia. A radical new intervention, the volume will be indispensable to scholars and researchers of history, sociology and social anthropology, politics, and South Asian studies.