We introduce DAiSEE, the first multi-label video classification dataset comprising of 9068 video snippets captured from 112 users for recognizing the user affective states of boredom, confusion, engagement, and frustration in the wild. The dataset has four levels of labels namely - very low, low, high, and very high for each of the affective states, which are crowd annotated and correlated with a gold standard annotation created using a team of expert psychologists. We have also established benchmark results on this dataset using state-of-the-art video classification methods that are available today. We believe that DAiSEE will provide the research community with challenges in feature extraction, context-based inference, and development of suitable machine learning methods for related tasks, thus providing a springboard for further research. The dataset is available for download at this https URL.

We introduce DriveIndia, a large-scale object detection dataset purpose-built to capture the complexity and unpredictability of Indian traffic environments. The dataset contains 66,986 high-resolution images annotated in YOLO format across 24 traffic-relevant object categories, encompassing diverse conditions such as varied weather (fog, rain), illumination changes, heterogeneous road infrastructure, and dense, mixed traffic patterns and collected over 120+ hours and covering 3,400+ kilometers across urban, rural, and highway routes. DriveIndia offers a comprehensive benchmark for real-world autonomous driving challenges. We provide baseline results using state-of-the-art YOLO family models, with the top-performing variant achieving a mAP50 of 78.7%. Designed to support research in robust, generalizable object detection under uncertain road conditions, DriveIndia will be publicly available via the TiHAN-IIT Hyderabad dataset repository this https URL (Terrestrial Datasets -> Camera Dataset).

This paper develops an agent-based automated fact-checking approach for detecting misinformation. We demonstrate that combining a powerful LLM agent, which does not have access to the internet for searches, with an online web search agent yields better results than when each tool is used independently. Our approach is robust across multiple models, outperforming alternatives and increasing the macro F1 of misinformation detection by as much as 20 percent compared to LLMs without search. We also conduct extensive analyses on the sources our system leverages and their biases, decisions in the construction of the system like the search tool and the knowledge base, the type of evidence needed and its impact on the results, and other parts of the overall process. By combining strong performance with in-depth understanding, we hope to provide building blocks for future search-enabled misinformation mitigation systems.

The growing demand for large-scale GPU clusters in distributed model training presents a significant barrier to innovation, particularly in model optimization, performance tuning, and system-level enhancements. To address this challenge, LLM training simulators are employed to estimate training time and guide design decisions. However, the state-of-the-art LLM training simulators assume homogeneous compute and network infrastructure. In practice, device heterogeneity is inevitable due to resource sharing in cloud environments, frequent shifts in device generations, and inherent intra-chip interconnect heterogeneity. To address the gap between state-of-the-art and practical requirements, we propose the design of a heterogeneity-aware distributed LLM simulator capable of predicting training time while enabling abstractions to specify custom configurations for device groups and device-to-parallelism mapping. We present the design requirements and challenges in building a heterogeneity-aware distributed ML training simulator, and design components such as non-uniform workload partitioning. Our initial simulation results demonstrate the impact of heterogeneity on the model computation and communication time.

Large Language Models (LLMs) are prevalent in modern applications but often memorize training data, leading to privacy breaches and copyright issues. Existing research has mainly focused on posthoc analyses, such as extracting memorized content or developing memorization metrics, without exploring the underlying architectural factors that contribute to memorization. In this work, we investigate memorization from an architectural lens by analyzing how attention modules at different layers impact its memorization and generalization performance. Using attribution techniques, we systematically intervene in the LLM architecture by bypassing attention modules at specific blocks while keeping other components like layer normalization and MLP transformations intact. We provide theorems analyzing our intervention mechanism from a mathematical view, bounding the difference in layer outputs with and without our attributions. Our theoretical and empirical analyses reveal that attention modules in deeper transformer blocks are primarily responsible for memorization, whereas earlier blocks are crucial for the models generalization and reasoning capabilities. We validate our findings through comprehensive experiments on different LLM families (Pythia and GPTNeo) and five benchmark datasets. Our insights offer a practical approach to mitigate memorization in LLMs while preserving their performance, contributing to safer and more ethical deployment in real world applications.

As photometric surveys reach unprecedented statistical precision, systematic uncertainties increasingly dominate large-scale structure probes relying on galaxy number density. Defining the final survey footprint is critical, as it excludes regions affected by artefacts or suboptimal observing conditions. For galaxy clustering, spatially varying observational systematics, such as seeing, are a leading source of bias. Template maps of contaminants are used to derive spatially dependent corrections, but extreme values may fall outside the applicability range of mitigation methods, compromising correction reliability. The complexity and accuracy of systematics modelling depend on footprint conservativeness, with aggressive masking enabling simpler, robust mitigation. We present a unified approach to define the DES Year 6 joint footprint, integrating observational systematics templates and artefact indicators that degrade mitigation performance. This removes extreme values from an initial seed footprint, leading to the final joint footprint. By evaluating the DES Year 6 lens sample MagLim++ plus plus on this footprint, we enhance the Iterative Systematics Decontamination (ISD) method, detecting non-linear systematic contamination and improving correction accuracy. While the mask's impact on clustering is less significant than systematics decontamination, it remains non-negligible, comparable to statistical uncertainties in certain w(theta) scales and redshift bins. Supporting coherent analyses of galaxy clustering and cosmic shear, the final footprint spans 4031.04 deg2, setting the basis for DES Year 6 1x2pt, 2x2pt, and 3x2pt analyses. This work highlights how targeted masking strategies optimise the balance between statistical power and systematic control in Stage-III and -IV surveys.

A vast amount of textual data is added to the internet daily, making utilization and interpretation of such data difficult and cumbersome. As a result, automatic text summarization is crucial for extracting relevant information, saving precious reading time. Although many transformer-based models excel in summarization, they are constrained by their input size, preventing them from processing texts longer than their context size. This study introduces three novel algorithms that allow any LLM to efficiently overcome its input size limitation, effectively utilizing its full potential without any architectural modifications. We test our algorithms on texts with more than 70,000 words, and our experiments show a significant increase in BERTScore with competitive ROUGE scores.

We perform a search for stellar streams around the Milky Way using the first three years of multi-band optical imaging data from the Dark Energy Survey (DES). We use DES data covering $\sim 5000$ sq. deg. to a depth of g > 23.5 with a relative photometric calibration uncertainty of < 1 \%. This data set yields unprecedented sensitivity to the stellar density field in the southern celestial hemisphere, enabling the detection of faint stellar streams to a heliocentric distance of $\sim 50$ kpc. We search for stellar streams using a matched-filter in color-magnitude space derived from a synthetic isochrone of an old, metal-poor stellar population. Our detection technique recovers four previously known thin stellar streams: Phoenix, ATLAS, Tucana III, and a possible extension of Molonglo. In addition, we report the discovery of eleven new stellar streams. In general, the new streams detected by DES are fainter, more distant, and lower surface brightness than streams detected by similar techniques in previous photometric surveys. As a by-product of our stellar stream search, we find evidence for extra-tidal stellar structure associated with four globular clusters: NGC 288, NGC 1261, NGC 1851, and NGC 1904. The ever-growing sample of stellar streams will provide insight into the formation of the Galactic stellar halo, the Milky Way gravitational potential, as well as the large- and small-scale distribution of dark matter around the Milky Way.

For text-based AI systems to interact in the real world, causal reasoning is an essential skill. Since active interventions are costly, we study to what extent a system can learn causal reasoning from symbolic demonstrations of causal axioms. Specifically, we present an axiomatic training method where the system learns from multiple demonstrations of a causal axiom (or rule), rather than incorporating the axiom as an inductive bias or inferring it from data values. A key question is whether the system would learn to generalize from the axiom demonstrations to more complex scenarios. Our results, based on applying axiomatic training to learn the transitivity axiom and d-separation rule, indicate that such generalization is possible. To avoid data contamination issues, we start with a 67 million parameter transformer model and train it from scratch. On both tasks, we find that a model trained on linear causal chains (along with some noisy variations) can generalize well to complex graphs, including longer causal chains, causal chains with reversed order, and graphs with this http URL handle diverse text inputs, the same method is extended to finetune language models. Finetuning Llama-3-8B-Instruct model on our axiomatic data leads to significant gains on causal benchmarks such as Corr2Cause and CLEAR, in some cases providing state-of-the-art performance surpassing GPT-4.

Recent efforts in multi-domain learning for semantic segmentation attempt to learn multiple geographical datasets in a universal, joint model. A simple fine-tuning experiment performed sequentially on three popular road scene segmentation datasets demonstrates that existing segmentation frameworks fail at incrementally learning on a series of visually disparate geographical domains. When learning a new domain, the model catastrophically forgets previously learned knowledge. In this work, we pose the problem of multi-domain incremental learning for semantic segmentation. Given a model trained on a particular geographical domain, the goal is to (i) incrementally learn a new geographical domain, (ii) while retaining performance on the old domain, (iii) given that the previous domain's dataset is not accessible. We propose a dynamic architecture that assigns universally shared, domain-invariant parameters to capture homogeneous semantic features present in all domains, while dedicated domain-specific parameters learn the statistics of each domain. Our novel optimization strategy helps achieve a good balance between retention of old knowledge (stability) and acquiring new knowledge (plasticity). We demonstrate the effectiveness of our proposed solution on domain incremental settings pertaining to real-world driving scenes from roads of Germany (Cityscapes), the United States (BDD100k), and India (IDD).

With the advancements made in deep learning, computer vision problems like object detection and segmentation have seen a great improvement in performance. However, in many real-world applications such as autonomous driving vehicles, the risk associated with incorrect predictions of objects is very high. Standard deep learning models for object detection such as YOLO models are often overconfident in their predictions and do not take into account the uncertainty in predictions on out-of-distribution data. In this work, we propose an efficient and effective approach to model uncertainty in object detection and segmentation tasks using Monte-Carlo DropBlock (MC-DropBlock) based inference. The proposed approach applies drop-block during training time and test time on the convolutional layer of the deep learning models such as YOLO. We show that this leads to a Bayesian convolutional neural network capable of capturing the epistemic uncertainty in the model. Additionally, we capture the aleatoric uncertainty using a Gaussian likelihood. We demonstrate the effectiveness of the proposed approach on modeling uncertainty in object detection and segmentation tasks using out-of-distribution experiments. Experimental results show that MC-DropBlock improves the generalization, calibration, and uncertainty modeling capabilities of YOLO models in object detection and segmentation.

Infrared divergences in the perturbative expansion of gauge theory amplitudes and cross sections have been a focus of theoretical investigations for almost a century. New insights still continue to emerge, as higher perturbative orders are explored, and high-precision phenomenological applications demand an ever more refined understanding. This review aims to provide a pedagogical overview of the subject. We briefly cover some of the early historical results, we provide some simple examples of low-order applications in the context of perturbative QCD, and discuss the necessary tools to extend these results to all perturbative orders. Finally, we describe recent developments concerning the calculation of soft anomalous dimensions in multi-particle scattering amplitudes at high orders, and we provide a brief introduction to the very active field of infrared subtraction for the calculation of differential distributions at colliders.

Tokenization is a crucial step in NLP, especially with the rise of large language models (LLMs), impacting downstream performance, computational cost, and efficiency. Existing LLMs rely on the classical Byte-pair Encoding (BPE) algorithm for subword tokenization that greedily merges frequent character bigrams, often leading to segmentation that does not align with linguistically meaningful units. To address this, we propose morphology-aware segmentation as a pre-tokenization step before applying BPE. To facilitate morphology-aware segmentation, we create a novel dataset for Hindi and Marathi, incorporating sandhi splitting to enhance the subword tokenization. Experiments on downstream tasks show that morphologically grounded tokenization improves machine translation and language modeling performance. Additionally, to handle the dependent vowels common in syllable-based writing systems used by Indic languages, we propose Constrained BPE (CBPE), an extension to the standard BPE algorithm incorporating script-specific constraints. In particular, CBPE handles dependent vowels to form a cohesive unit with other characters instead of occurring as a single unit. Our results show that CBPE achieves a 1.68\% reduction in fertility scores while maintaining comparable or improved downstream performance in machine translation and language modeling, offering a computationally efficient alternative to standard BPE. Moreover, to evaluate segmentation across different tokenization algorithms, we introduce a new human evaluation metric, \textit{EvalTok}, enabling more human-grounded assessment.

Binary similarity involves determining whether two binary programs exhibit similar functionality, often originating from the same source code. In this work, we propose VexIR2Vec, an approach for binary similarity using VEX-IR, an architecture-neutral Intermediate Representation (IR). We extract the embeddings from sequences of basic blocks, termed peepholes, derived by random walks on the control-flow graph. The peepholes are normalized using transformations inspired by compiler optimizations. The VEX-IR Normalization Engine mitigates, with these transformations, the architectural and compiler-induced variations in binaries while exposing semantic similarities. We then learn the vocabulary of representations at the entity level of the IR using the knowledge graph embedding techniques in an unsupervised manner. This vocabulary is used to derive function embeddings for similarity assessment using VexNet, a feed-forward Siamese network designed to position similar functions closely and separate dissimilar ones in an n-dimensional space. This approach is amenable for both diffing and searching tasks, ensuring robustness against Out-Of-Vocabulary (OOV) issues. We evaluate VexIR2Vec on a dataset comprising 2.7M functions and 15.5K binaries from 7 projects compiled across 12 compilers targeting x86 and ARM architectures. In diffing experiments, VexIR2Vec outperforms the nearest baselines by $40\%$, $18\%$, $21\%$, and $60\%$ in cross-optimization, cross-compilation, cross-architecture, and obfuscation settings, respectively. In the searching experiment, VexIR2Vec achieves a mean average precision of $0.76$, outperforming the nearest baseline by $46\%$. Our framework is highly scalable and is built as a lightweight, multi-threaded, parallel library using only open-source tools. VexIR2Vec is $3.1$-$3.5 \times$ faster than the closest baselines and orders-of-magnitude faster than other tools.

Observations of density variations in stellar streams are a promising probe of low-mass dark matter substructure in the Milky Way. However, survey systematics such as variations in seeing and sky brightness can also induce artificial fluctuations in the observed densities of known stellar streams. These variations arise because survey conditions affect both object detection and star-galaxy misclassification rates. To mitigate these effects, we use Balrog synthetic source injections in the Dark Energy Survey (DES) Y3 data to calculate detection rate variations and classification rates as functions of survey properties. We show that these rates are nearly separable with respect to survey properties and can be estimated with sufficient statistics from the synthetic catalogs. Applying these corrections reduces the standard deviation of relative detection rates across the DES footprint by a factor of five, and our corrections significantly change the inferred linear density of the Phoenix stream when including faint objects. Additionally, for artificial streams with DES like survey properties we are able to recover density power spectra with reduced bias. We also find that uncorrected power-spectrum results for LSST-like data can be around five times more biased, highlighting the need for such corrections in future ground based surveys.

The evolution of digital image manipulation, particularly with the advancement of deep generative models, significantly challenges existing deepfake detection methods, especially when the origin of the deepfake is obscure. To tackle the increasing complexity of these forgeries, we propose \textbf{Wavelet-CLIP}, a deepfake detection framework that integrates wavelet transforms with features derived from the ViT-L/14 architecture, pre-trained in the CLIP fashion. Wavelet-CLIP utilizes Wavelet Transforms to deeply analyze both spatial and frequency features from images, thus enhancing the model's capability to detect sophisticated deepfakes. To verify the effectiveness of our approach, we conducted extensive evaluations against existing state-of-the-art methods for cross-dataset generalization and detection of unseen images generated by standard diffusion models. Our method showcases outstanding performance, achieving an average AUC of 0.749 for cross-data generalization and 0.893 for robustness against unseen deepfakes, outperforming all compared methods. The code can be reproduced from the repo: \url{this https URL}

We study the unbalanced optimal transport (UOT) problem, where the marginal constraints are enforced using Maximum Mean Discrepancy (MMD) regularization. Our work is motivated by the observation that the literature on UOT is focused on regularization based on $\phi$-divergence (e.g., KL divergence). Despite the popularity of MMD, its role as a regularizer in the context of UOT seems less understood. We begin by deriving a specific dual of MMD-regularized UOT (MMD-UOT), which helps us prove several useful properties. One interesting outcome of this duality result is that MMD-UOT induces novel metrics, which not only lift the ground metric like the Wasserstein but are also sample-wise efficient to estimate like the MMD. Further, for real-world applications involving non-discrete measures, we present an estimator for the transport plan that is supported only on the given ($m$) samples. Under certain conditions, we prove that the estimation error with this finitely-supported transport plan is also $\mathcal{O}(1/\sqrt{m})$. As far as we know, such error bounds that are free from the curse of dimensionality are not known for $\phi$-divergence regularized UOT. Finally, we discuss how the proposed estimator can be computed efficiently using accelerated gradient descent. Our experiments show that MMD-UOT consistently outperforms popular baselines, including KL-regularized UOT and MMD, in diverse machine learning applications.