Faisal Hossain Raquib, Akm Moshiur Rahman Mazumder, Md Fahim, Md Tahmid Hasan Fuad, Md Farhan Ishmam, Faria Sultana, M Ashraful Amin, Amin Ahsan Ali, Akmmahbubur Rahman. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Pre-trained vision-language models (VLMs) such as CLIP exhibit strong generalization but struggle with few-shot adaptation due to the trade-off between gaining task-specific knowledge and preserving general performance. While multimodal adapters add trainable modules that improve alignment and excel in few-shot generality, they greatly increase the trainable parameter count while relying heavily on the prior layer’s frozen representation. Addressing these limitations, we introduce Recurrent Multi-Modal Adapter (R-MMA), a lightweight and efficient adapter that uses self-attention to compute a unified latent representation with a single set of shared adapter weights. Our attention-based alignment harmonizes the adapter outputs with the frozen encoder features before fusing the modalities, ensuring better preservation of pre-trained representations and cross-modal consistency. Our experiments show that R-MMA achieves state-of-the-art performance on most datasets for base-to-novel generalization, cross-dataset evaluation, and domain generalization, under few-shot settings. Our approach also achieves one of the highest forms of parameter efficiency with only a few trainable weight matrices for the whole network, regardless of its depth. Our code is available at: https://github.com/farhanishmam/R-MMA.

We introduce a novel machine learning dataset tailored for the classification of bent radio active galactic nuclei (AGN) in astronomical observations. Bent radio AGN, distinguished by their curved jet structures, provide critical insights into galaxy cluster dynamics, interactions within the intracluster medium, and the broader physics of AGN. Despite their astrophysical significance, the classification of bent radio AGN remains a challenge due to the scarcity of specialized datasets and benchmarks. To address this, we present a dataset, derived from a well-recognized radio astronomy survey, that is designed to support the classification of NAT (Narrow-Angle Tail) and WAT (Wide-Angle Tail) categories, along with detailed data processing steps. We further evaluate the performance of state-of-the-art deep learning models on the dataset, including Convolutional Neural Networks (CNNs), and transformer-based architectures. Our results demonstrate the effectiveness of advanced machine learning models in classifying bent radio AGN, with ConvNeXT achieving the highest F1-scores for both NAT and WAT sources. By sharing this dataset and benchmarks, we aim to facilitate the advancement of research in AGN classification, galaxy cluster environments and galaxy evolution. The source code is available at: https://github.com/mirsazzathossain/RGC-Bent

Land Use Land Cover (LULC) mapping using deep learning significantly enhances the reliability of LULC classification, aiding in understanding geography, socioeconomic conditions, poverty levels, and urban sprawl. However, the scarcity of annotated satellite data, especially in South/East Asian developing countries, poses a major challenge due to limited funding, diverse infrastructures, and dense populations. In this work, we introduce the BD Open LULC Map (BOLM), providing pixel-wise LULC annotations across eleven classes (e.g., Farmland, Water, Forest, Urban Structure, Rural Built- Up) for Dhaka metropolitan city and its surroundings using high-resolution Bing satellite imagery (2.22 m/pixel). BOLM spans 4,392 km2(891 million pixels), with ground truth validated through a three-stage process involving GIS experts. We benchmark LULC segmentation using DeepLab V3+ across five major classes and compare performance on Bing and Sentinel-2A imagery. BOLM aims to support reliable deep models and domain adaptation tasks, addressing critical LULC dataset gaps in South/East Asia.

Federated Learning (FL) enables collaborative model training across distributed clients without sharing raw data, making it ideal for privacy-sensitive applications. However, FL models often suffer performance degradation due to distribution shifts between training and deployment. Test-Time Adaptation (TTA) offers a promising solution by allowing models to adapt using only test samples. However, existing TTA methods in FL face challenges such as computational overhead, privacy risks from feature sharing, and scalability concerns due to memory constraints. To address these limitations, we propose Federated Continual Test-Time Adaptation (FedCTTA), a privacy-preserving and computationally efficient framework for federated adaptation. Unlike prior methods that rely on sharing local feature statistics, FedCTTA avoids direct feature exchange by leveraging similarity-aware aggregation based on model output distributions over randomly generated noise samples. This approach ensures adaptive knowledge sharing while preserving data privacy. Furthermore, FedCTTA minimizes the entropy at each client for continual adaptation, enhancing the model’s confidence in evolving target distributions. Our method eliminates the need for server-side training during adaptation and maintains a constant memory footprint, making it scalable even as the number of clients or training rounds increases. Extensive experiments show that FedCTTA surpasses existing methods across diverse temporal and spatial heterogeneity scenarios.

Social media users express their feelings, experiences, ideas, and stories with little or no regard for the conventions of traditional grammar. Online discourse, by its very nature, is rife with transliterated text along with code-mixing and code-switching. Transliteration is heavily featured due to the ease of inputting romanized text with standard keyboards over native scripts. Due to its ubiquity, it is a critical area of study to ensure NLP models perform well in real-world scenarios. In this paper, we analyze the performance of various language models, Tiny Large Language models, TF-IDF and Bag-of-Words feature extraction-based classical ML models, as well as zero-shot classification with ChatGPT on romanized/transliterated social media text. We chose the tasks of sentiment analysis and offensive language identification and we carried out experiments for three different languages, namely Bangla, Hindi, and Arabic, for six datasets. To our surprise, we discovered across multiple datasets that the non-neural methods perform very competitively with fine-tuned transformer-based mono/multilingual language models, tiny large language models, and ChatGPT for classification tasks in transliterated text. These classical models train in seconds using only a fraction of the computing power, and thus the carbon footprint, required by language models. We demonstrate TF-IDF and BoW-based classifiers achieve performance within around 3% of fine-tuned LMs and could thus be considered as a strong baseline for transliterated text-based NLP tasks. Additionally, we investigated various mitigation strategies such as translation and augmentation via the use of ChatGPT, as well as Masked Language Modelling to dataset-specific pretraining for language models. Depending on the dataset and language, employing those mitigation techniques yields a 2-3% further improvement in accuracy and macro-F1 above baseline.

In recent years, significant progress has been made in image super-resolution through the use of large-scale models. However, the efficacy of these models comes at the cost of their substantial size, posing challenges and limitations when deploying them on resource-constrained devices. Despite their remarkable performance, the feasibility of employing such models on low-end devices has remained a contentious topic. In light of this, our research introduces a lightweight approach to image super-resolution, leveraging a simple recurrent neural network architecture consisting of a recurrent convolution block. Our proposed model uses less than 75k parameters, which is 10 times fewer than the state-of-the-art transformer-based super-resolution model. Despite its small size, the proposed model performs well in image super-resolution tasks both visually and quantitatively. Our work presents a promising direction for addressing the difficulty of deploying efficient super-resolution models on resource-limited devices.

Emotion recognition is an exacting task due to the complexity and diversity of human emotions, as well as their tendency to be open to multiple interpretations. Despite these challenges, emotion recognition from textual data has achieved promising results in recent years for the English language. Numerous studies have been conducted on this topic, leading to significant advancements in various fields, including business, politics, education, research, and healthcare. However, accurate emotion recognition from Bengali (Bangla) text remains difficult due to limited resources available for the Bangla language. Additionally, the available datasets for Bengali text are often poorly annotated or imbalanced, which significantly hampers the performance of text classifiers and causes overfitting. Data augmentation can help address this issue. By addressing class imbalance through data augmentation via back-translation, we analyzed and evaluated the impact of augmentation on both real and synthetic datasets. We used transformer-based models, Multi-BERT and BanglaBERT. Data augmentation considerably boosts the accuracy by 8% and 6% for MultiBERT and BanglaBERT respectively, along with a 13% increase in F1-score for both the models. An ablation study also evaluated the effects of hyperparameters on our models.

While describing spatiotemporal events in natural language, video captioning models mostly rely on the en-coder’s latent visual representation. Recent progress on the encoder-decoder model attends encoder features mainly in linear interaction with the decoder. However, growing model complexity for visual data encourages more explicit feature interaction for fine-grained information, which is currently absent in the video captioning domain. Moreover, feature aggregations methods have been used to un-veil richer visual representation, either by the concatenation or using a linear layer. Though feature sets for a video semantically overlap to some extent, these approaches result in objective mismatch and feature redundancy. In addition, diversity in captions is a fundamental component of expressing one event from several meaningful perspectives, currently missing in the temporal, i.e., video captioning domain. To this end, we propose Variational Stacked Local Attention Network (VSLAN), which exploits low-rank bilinear pooling for self-attentive feature interaction and stacking multiple video feature streams in a discount fashion. Each feature stack’s learned attributes contribute to our proposed diversity encoding module, followed by the decoding query stage to facilitate end-to-end diverse and natural captions without any explicit supervision on attributes. We evaluate VSLAN on MSVD and MSR-VTT datasets in terms of syntax and diversity. The CIDEr score of VSLAN outperforms current off-the-shelf methods by 7.8% on MSVD and 4.5% on MSR-VTT, respectively. On the same datasets, VSLAN achieves competitive results in caption diversity metrics.

Women in the global south often seek justice to their online harassment through unveiling the harassers and the screenshots of their sent harassment texts and visual contents before the relevant authorities. Nevertheless, such evidence is often challenged for their authenticity. Our survey (n=91) and interview (n=43) with Bangladeshi online gender harassment victims revealed the depth of the problem, and we set design goals to collect evidence from Facebook Messenger with ensured authenticity. Building on the ‘shame-based model’ of gender justice [12], we designed ‘Unmochon’, a tool that captures authentic evidence and shares with victims’ intended group. Our user-study (n=48) revealed that diminishing authenticity problem may still leave the victim and online gender justice entangled with mob-sentiment, hegemonic legal consciousness, and several privacy aspects. Our findings open up a new discussion on how HCI-design should address online gender justice in such a complex social setting.

High-resolution image segmentation remains challenging and error-prone due to the enormous size of intermediate feature maps. Conventional methods avoid this problem by using patch based approaches where each patch is segmented independently. However, independent patch segmentation induces errors, particularly at the patch boundary due to the lack of contextual information in very high-resolution images where the patch size is much smaller compared to the full image. To overcome these limitations, in this paper, we propose a novel framework to segment a particular patch by incorporating contextual information from its neighboring patches. This allows the segmentation network to see the target patch with a wider field of view without the need of larger feature maps. Comparative analysis from a number of experiments shows that our proposed framework is able to segment high resolution images with significantly improved mean Intersection over Union and overall accuracy.

Research in deep learning models to forecast traffic intensities has gained great attention in recent years due to their capability to capture the complex spatio-temporal relationships within the traffic data. However, most state-of-the-art approaches have designed spatial-only (e.g. Graph Neural Networks) and temporal-only (e.g. Recurrent Neural Networks) modules to separately extract spatial and temporal features. However, we argue that it is less effective to extract the complex spatiotemporal relationship with such factorized modules. Besides, most existing works predict the traffic intensity of a particular time interval only based on the traffic data of the previous one hour of that day. And thereby ignores the repetitive daily/weekly pattern that may exist in the last hour of data. Therefore, we propose a Unified Spatio-Temporal Graph Convolution Network (USTGCN) for traffic forecasting that performs both spatial and temporal aggregation through direct information propagation across different timestamp nodes with the help of spectral graph convolution on a spatio-temporal graph. Furthermore, it captures historical daily patterns in previous days and current-day patterns in current-day traffic data. Finally, we validate our work's effectiveness through experimental analysis11Code is available at github.com/AmitRoy7781/USTGCN, which shows that our model USTGCN can outperform state-of-the-art performances in three popular benchmark datasets from the Performance Measurement System (PeMS). Moreover, the training time is reduced significantly with our proposed USTGCN model.

Graph Neural Networks (GNNs) learn low dimensional representations of nodes by aggregating information from their neighborhood in graphs. However, traditional GNNs suffer from two fundamental shortcomings due to their local (l-hop neighborhood) aggregation scheme. First, not all nodes in the neighborhood carry relevant information for the target node. Since GNNs do not exclude noisy nodes in their neighborhood, irrelevant information gets aggregated, which reduces the quality of the representation. Second, traditional GNNs also fail to capture long-range non-local dependencies between nodes. To address these limitations, we exploit mutual information (MI) to define two types of neighborhood, 1) Local Neighborhood where nodes are densely connected within a community and each node would share higher MI with its neighbors, and 2) Non-Local Neighborhood where MI-based node clustering is introduced to assemble informative but graphically distant nodes in the same cluster. To generate node presentations, we combine the embeddings generated by bi-level aggregation - local aggregation to aggregate features from local neighborhoods to avoid noisy information and non-local aggregation to aggregate features from non-local neighborhoods. Furthermore, we leverage self-supervision learning to estimate MI with few labeled data. Finally, we show that our model significantly outperforms the state-of-the-art methods in a wide range of assortative and disassortative graphs11Source Code at: https://github.com/forkkr/LnL-GNN.

Graph pooling is an essential ingredient of Graph Neural Networks (GNNs) in graph classification and regression tasks. For these tasks, different pooling strategies have been proposed to generate a graph-level representation by downsampling and summarizing nodes' features in a graph. However, most existing pooling methods are unable to capture distinguishable structural information effectively. Besides, they are prone to adversarial attacks. In this work, we propose a novel pooling method named as HIBPool where we leverage the Information Bottleneck (IB) principle that optimally balances the expressiveness and robustness of a model to learn representations of input data. Furthermore, we introduce a novel structure-aware Discriminative Pooling Readout (DiP-Readout) function to capture the informative local subgraph structures in the graph. Finally, our experimental results show that our model significantly outperforms other state-of-art methods on several graph classification benchmarks and more resilient to feature-perturbation attack than existing pooling methods11Source code at: https://github.com/forkkr/HIBPool.

Wearable sensor based human activity recognition is a challenging problem due to difficulty in modeling spatial and temporal dependencies of sensor signals. Recognition models in closed-set assumption are forced to yield members of known activity classes as prediction. However, activity recognition models can encounter an unseen activity due to body-worn sensor malfunction or disability of the subject performing the activities. This problem can be addressed through modeling solution according to the assumption of open-set recognition. Hence, the proposed self attention based approach combines data hierarchically from different sensor placements across time to classify closed-set activities and it obtains notable performance improvement over state-of-the-art models on five publicly available datasets. The decoder in this autoencoder architecture incorporates self-attention based feature representations from encoder to detect unseen activity classes in open-set recognition setting. Furthermore, attention maps generated by the hierarchical model demonstrate explainable selection of features in activity recognition. We conduct extensive leave one subject out validation experiments that indicate significantly improved robustness to noise and subject specific variability in body-worn sensor signals. The source code is available at: github.com/saif-mahmud/hierarchical-attention-HAR

Human Activity Recognition from body-worn sensor data poses an inherent challenge in capturing spatial and temporal dependencies of time-series signals. In this regard, the existing recurrent or convolutional or their hybrid models for activity recognition struggle to capture spatio-temporal context from the feature space of sensor reading sequence. To address this complex problem, we propose a self-attention based neural network model that foregoes recurrent architectures and utilizes different types of attention mechanisms to generate higher dimensional feature representation used for classification. We performed extensive experiments on four popular publicly available HAR datasets: PAMAP2, Opportunity, Skoda and USC-HAD. Our model achieve significant performance improvement over recent state-of-the-art models in both benchmark test subjects and Leave-one-subject-out evaluation. We also observe that the sensor attention maps produced by our model is able capture the importance of the modality and placement of the sensors in predicting the different activity classes.

The advancement of deep learning technology has enabled us to develop systems that outperform any other classification technique. However, success of any empirical system depends on the quality and diversity of the data available to train the proposed system. In this research, we have carefully accumulated a relatively challenging dataset that contains images collected from various sources for three different disasters: fire, water and land. Besides this, we have also collected images for various damaged infrastructure due to natural or man made calamities and damaged human due to war or accidents. We have also accumulated image data for a class named non-damage that contains images with no such disaster or sign of damage in them. There are 13,720 manually annotated images in this dataset, each image is annotated by three individuals. We are also providing discriminating image class information annotated manually with bounding box for a set of 200 test images. Images are collected from different news portals, social media, and standard datasets made available by other researchers. A three layer attention model (TLAM) is trained and average five fold validation accuracy of 95.88% is achieved. Moreover, on the 200 unseen test images this accuracy is 96.48%. We also generate and compare attention maps for these test images to determine the characteristics of the trained attention model.

Pre-processing, Feature Extraction, Feature Selection and Classification are the four sub modules of the Signal Processing module of a typical BCI system. Pattern recognition is mainly involved in this Signal Processing module and in this paper, we experimented with different state-of-the-art algorithms for each of these submodules on two separate datasets we acquired using Emotiv EPOC and the Muse headband from 38 college-aged young adults. For our experiment, we used two artefact removal techniques, namely Stationary Wavelet Transform (SWT) based denoising technique and an extended SWT technique (SWTSD). We found SWTSD improves average classification accuracy up to 7.2 % and performs better than SWT. However, that does not state that SWTSD will outperform SWT when implemented on other BCI paradigms or on other EEG-based applications. In our study, the highest average accuracy achieved by the data of the Muse headband and Emotiv EPOC were 77.7% and 66.7% respectively and from our results we conclude that, the performance of different BCI systems depends on several different factors including artefact removal techniques, filters, feature extraction and selection algorithms, classifiers, etc. and appropriate choice and usage of such methods can have a significant positive impact on the end results.

This paper presents a framework for building diagnostic imaging teaching and learning facility for entry level medical students of Bangladesh. Initially we demonstrate an X-Ray image analysis and retrieval system that will work as one of the main component in this system. This web based system has three modes. First is the annotation mode where an expert radiologist manually performs annotation of raw x-ray images. To aid the annotation process proposed model proposes a manual and a semi-auto segmentation tool in identifying the region of interests (ROI) in the X-Ray images. Image Retrieval in Medical Applications (IRMA) structure has been used for the annotating the ROIs. In the learning mode, students can retrieve images from the database created by expert radiologists. We proposed information retrieval techniques to find x-ray images of interest. We have used text based and content based search methods which is based on term frequency–inverse document frequency (tf-idf), and Gabor filter respectively.

This paper examines the classification capability of different Gabor representations for human face recognition. Usually, Gabor filter responses for eight orientations and five scales for each orientation are calculated and all 40 basic feature vectors are concatenated to assemble the Gabor feature vector. This work explores 70 different Gabor feature vector extraction techniques for face recognition. The main goal is to determine the characteristics of the 40 basic Gabor feature vectors and to devise a faster Gabor feature extraction method. Among all the 40 basic Gabor feature representations the filter responses acquired from the largest scale at smallest relative orientation change (with respect to face) shows the highest discriminating ability for face recognition while classification is performed using three classification methods: probabilistic neural networks (PNN), support vector machines (SVM) and decision trees (DT). A 40 times faster summation based Gabor representation shows about 98% recognition rate while classification is performed using SVM. In this representation all 40 basic Gabor feature vectors are summed to form the summation based Gabor feature vector. In the experiment, a sixth order data tensor containing the basic Gabor feature vectors is constructed, for all the operations.

One of the important issues in image feature is the area of objects in the image. In this paper we propose an area finding algorithm to measure the area of an image object. This method finds the area by polygonal approximation of image objects. A sequential fine-tuning of this polygon is used to estimate the actual area from polygon area. This area modification sequence is used as a topological description of an image object and used as feature in image matching.