Adapting foundation models for specific purposes has become a standard approach to build machine learning systems for downstream applications. Yet, it is an open question which mechanisms take place during adaptation. Here we develop a new Sparse Autoencoder (SAE) for the CLIP vision transformer, named PatchSAE, to extract interpretable concepts at granular levels (e.g., shape, color, or semantics of an object) and their patch-wise spatial attributions. We explore how these concepts influence the model output in downstream image classification tasks and investigate how recent state-of-the-art prompt-based adaptation techniques change the association of model inputs to these concepts. While activations of concepts slightly change between adapted and non-adapted models, we find that the majority of gains on common adaptation tasks can be explained with the existing concepts already present in the non-adapted foundation model. This work provides a concrete framework to train and use SAEs for Vision Transformers and provides insights into explaining adaptation mechanisms.

Dataset bias, where data points are skewed to certain concepts, is ubiquitous in machine learning datasets. Yet, systematically identifying these biases is challenging without costly, fine-grained attribute annotations. We present ConceptScope, a scalable and automated framework for analyzing visual datasets by discovering and quantifying human-interpretable concepts using Sparse Autoencoders trained on representations from vision foundation models. ConceptScope categorizes concepts into target, context, and bias types based on their semantic relevance and statistical correlation to class labels, enabling class-level dataset characterization, bias identification, and robustness evaluation through concept-based subgrouping. We validate that ConceptScope captures a wide range of visual concepts, including objects, textures, backgrounds, facial attributes, emotions, and actions, through comparisons with annotated datasets. Furthermore, we show that concept activations produce spatial attributions that align with semantically meaningful image regions. ConceptScope reliably detects known biases (e.g., background bias in Waterbirds) and uncovers previously unannotated ones (e.g, co-occurring objects in ImageNet), offering a practical tool for dataset auditing and model diagnostics.

Interactive segmentation methods have been investigated to address the potential need for additional refinement in automatic segmentation via human-in-the-loop techniques. For accurate segmentation of 3D images, we propose Slice-and-Conquer, a novel planar-to-3D pipeline formulating volumetric mask construction into two stages: 1) 2D interactive segmentation and 2) guided 3D segmentation. Specifically, the first stage enables users to focus on a single 2D slice and provides the corresponding 2D prediction results as strong shape priors. Taking the planar guidance, an accurate 3D mask can be constructed with minimal interactions. To support a flexible iterative refinement, our system recommends a next slice to annotate at the end of the second stage. Since volumetric segmentation can be completed by consecutively annotating a few recommended 2D slices, our method significantly reduces the cognitive burden of exploring volumetric space for users. Through extensive experiments on various datasets of 3D biomedical images, we demonstrate the effectiveness of the proposed pipeline.

The 5G Core networks (5GC) are witnessing a drastic increase in signaling traffic volume due to flat Service Based Architecture (SBA) with “any to any” Network Function (NF) communication. The increased signaling, if not managed properly, may hamper the timely delivery of control messages, degrading the user's quality of experience. The underlying reason for increased 5GC signaling is high inter-dependencies between NFs that result in long, complex, and sequential system procedures. Moreover, system procedures are not optimally designed for cloud-native SBA based 5GC. This paper analyzes the system procedures for the current 5G network in order to decouple the sequential procedures into small and independent transactions. Once decoupled, the control messages can be executed in parallel, significantly reducing the procedure completion time and improving compute resource utilization. Although we decouple a few control messages in 5G registration procedure, we are limited by the design of NFs, which introduces inter- Nfcoupling. Thus, architectural changes are imperative to enable loose coupling between NFs. We present an end-to-end SBA based network for 6G and propose enhanced system procedures that allow parallel execution of control messages. We perform simulations for the modified registration procedures, demonstrating approximately 13 % and 24 % decrease in procedure completion time for 5G and 6G, respectively.

In this paper, we propose a new user-plane (UP) network architecture for 6G cellular networks, where there is no distinction between radio access network (RAN) and core network (CN). We define a new network function (NF), called merged-UP (M-UP), where the UP functions of both RAN and CN sides are merged, thus enabling efficient protocol operations. We also propose an IP flow-based radio quality-of-service (QoS) control scheme to support 6G services with diverse QoS requirements. Based on our network testbed system, we show that the proposed architecture improves the system performance in terms of packet processing load and maximum throughput.

Network virtualization and cloudification are being actively introduced into mobile networks, but the current control-plane (CP) architecture, which distinguishes radio access network (RAN) from core network (CN), is inefficient in fully utilizing the flexibility provided by such technologies. In this paper, we propose a novel CP architecture, which is based on service-based interface (SBI) throughout the network without distinction between RAN and CN, thus reducing the delay and overhead of control signaling. The proposed architecture not only reduces the completion time and amount of signaling of the control procedures by up to 27% and 45%, respectively, but also provides a cloud-friendly environment in which slices and sub-networks with various characteristics can be easily operated by configuring the CP at will.

Abstract The majority of visualizations on the web are still stored as raster images, making them inaccessible to visually impaired users. We propose a deep‐neural‐network‐based approach that automatically recognizes key elements in a visualization, including a visualization type, graphical elements, labels, legends, and most importantly, the original data conveyed in the visualization. We leverage such extracted information to provide visually impaired people with the reading of the extracted information. Based on interviews with visually impaired users, we built a Google Chrome extension designed to work with screen reader software to automatically decode charts on a webpage using our pipeline. We compared the performance of the back‐end algorithm with existing methods and evaluated the utility using qualitative feedback from visually impaired users.

Detecting anomalous events of a particular area in a timely manner is an important task. Geo-tagged social media data are useful resource for this task; however, the abundance of everyday language in them makes this task still challenging. To address such challenges, we present TopicOnTiles, a visual analytics system that can reveal information relevant to anomalous events in a multi-level tile-based map interface by using social media data. To this end, we adopt and improve a recently proposed topic modeling method that can extract spatio-temporally exclusive topics corresponding to a particular region and a time point. Furthermore, we utilize a tile-based map interface to efficiently handle large-scale data in parallel. Our user interface effectively highlights anomalous tiles using our novel glyph visualization that encodes the degree of anomaly computed by our exclusive topic modeling processes. To show the effectiveness of our system, we present several usage scenarios using real-world datasets as well as comprehensive user study results.

Understanding newly emerging events or topics associated with a particular region of a given day can provide deep insight on the critical events occurring in highly evolving metropolitan cities. We propose herein a novel topic modeling approach on text documents with spatio-temporal information (e.g., when and where a document was published) such as location-based social media data to discover prevalent topics or newly emerging events with respect to an area and a time point. We consider a map view composed of regular grids or tiles with each showing topic keywords from documents of the corresponding region. To this end, we present a tilebased spatio-temporally exclusive topic modeling approach called STExNMF, based on a novel nonnegative matrix factorization (NMF) technique. STExNMF mainly works based on the two following stages: (1) first running a standard NMF of each tile to obtain general topics of the tile and (2) running a spatiotemporally exclusive NMF on a weighted residual matrix. These topics likely reveal information on newly emerging events or topics of interest within a region. We demonstrate the advantages of our approach using the geo-tagged Twitter data of New York City. We also provide quantitative comparisons in terms of the topic quality, spatio-temporal exclusiveness, topic variation, and qualitative evaluations of our method using several usage scenarios. In addition, we present a fast topic modeling technique of our model by leveraging parallel computing.