Search

Agricultural landscapes are quite complex, especially in the Global South where fields are smaller, and agricultural practices are more varied. In this paper we report on our progress in digitizing the agricultural landscape (natural and man-made) in our study region of India. We use high resolution imagery and a UNet style segmentation model to generate the first of its kind national-scale multi-class panoptic segmentation output. Through this work we have been able to identify individual fields across 151.7M hectares, and delineating key features such as water resources and vegetation. We share how this output was validated by our team and externally by downstream users, including some sample use cases that can lead to targeted data driven decision making. We believe this dataset will contribute towards digitizing agriculture by generating the foundational baselayer., Comment: 34 pages, 7 tables, 15 figs

We report measurements of the absolute branching fractions $\mathcal{B}(B_s^0 \to D_s^{\pm} X)$, $\mathcal{B}(B_s^0 \to D^0/\bar{D}^0 X)$, and $\mathcal{B}(B_s^0 \to D^{\pm} X)$, where the latter is measured for the first time. The results are based on a 121.4\,fb$^{-1}$ data sample collected at the $\Upsilon(10860)$ resonance by the Belle detector at the KEKB asymmetric-energy $e^+ e^-$ collider. We reconstruct one $B_s^0$ meson in $e^+e^- \to \Upsilon(10860) \to B_s^{*} \bar{B}_s^{*}$ events and measure yields of $D_s^+$, $D^0$, and $D^+$ mesons in the rest of the event. We obtain $\mathcal{B}(B_s^0 \to D_s^{\pm} X) = (68.6 \pm 7.2 \pm 4.0)\%$, $\mathcal{B}(B_s^0 \to D^0/\bar{D}^0 X) = (21.5 \pm 6.1 \pm 1.8)\%$, and $\mathcal{B}(B_s^0 \to D^{\pm} X) = (12.6 \pm 4.6 \pm 1.3)\%$, where the first uncertainty is statistical and the second is systematic. Averaging with previous Belle measurements gives $\mathcal{B}(B_s^0 \to D_s^{\pm} X) = (63.4 \pm 4.5 \pm 2.2)\%$ and $\mathcal{B}(B_s^0 \to D^0/\bar{D}^0 X) = (23.9 \pm 4.1 \pm 1.8)\%$. For the $B_s^0$ production fraction at the $\Upsilon(10860)$, we find $f_s = (21.4^{+1.5}_{-1.7})\%$., Comment: 23 pages, 9 figures, submitted to JHEP

This paper presents Robust samplE-based coVarIance StEering (REVISE), a multi-query algorithm that generates robust belief roadmaps for dynamic systems navigating through spatially dependent disturbances modeled as a Gaussian random field. Our proposed method develops a novel robust sample-based covariance steering edge controller to safely steer a robot between state distributions, satisfying state constraints along the trajectory. Our proposed approach also incorporates an edge rewiring step into the belief roadmap construction process, which provably improves the coverage of the belief roadmap. When compared to state-of-the-art methods, REVISE improves median plan accuracy (as measured by Wasserstein distance between the actual and planned final state distribution) by 10x in multi-query planning and reduces median plan cost (as measured by the largest eigenvalue of the planned state covariance at the goal) by 2.5x in single-query planning for a 6DoF system. We will release our code at https://acl.mit.edu/REVISE/.

We present measurements of $B \to K{}^{*}(892)\gamma$ decays using $365\,{\rm fb}^{-1}$ of data collected from 2019 to 2022 by the Belle~II experiment at the SuperKEKB asymmetric-energy $e^+e^-$ collider. The data sample contains $(387 \pm 6) \times 10^6$ $B\overline{B}$ events. We measure branching fractions ($\mathcal{B}$) and $C\!P$ asymmetries ($\mathcal{A}_{C\!P}$) for both $B^{0}\to K{}^{*0}\gamma$ and $B^{+}\to K{}^{*+}\gamma$ decays. The difference in $C\!P$ asymmetries ($\Delta \mathcal{A}_{C\!P}$) and the isospin asymmetry ($\Delta_{0+}$) between these neutral and charged channels are also measured. We obtain the following branching fractions and $C\!P$ asymmetries: $\mathcal{B} (B^{0} \to K{}^{*0}\gamma) = (4.14 \pm 0.10 \pm 0.11 ) \times 10^{-5}$, $\mathcal{B} (B^{+} \to K{}^{*+}\gamma) = (4.02 \pm 0.13 \pm 0.13 )\times 10^{-5}$, $\mathcal{A}_{C\!P} (B^{0} \to K{}^{*0}\gamma) = (-3.3 \pm 2.3 \pm 0.4 )\%$, and $\mathcal{A}_{C\!P} (B^{+} \to K{}^{*+}\gamma) = (-0.7 \pm 2.9 \pm 0.6 )\%$. The measured difference in $C\!P$ asymmetries is $\Delta \mathcal{A}_{C\!P} = (+2.6 \pm 3.8 \pm 0.7 )\%$, and the measured isospin asymmetry is $\Delta_{0+} = (+5.0 \pm 2.0 \pm 1.5 )\%$. The first uncertainties listed are statistical and the second are systematic. These results are consistent with world-average values and theory predictions.

We study a setting where agents use no-regret learning algorithms to participate in repeated auctions. \citet{kolumbus2022auctions} showed, rather surprisingly, that when bidders participate in second-price auctions using no-regret bidding algorithms, no matter how large the number of interactions $T$ is, the runner-up bidder may not converge to bidding truthfully. Our first result shows that this holds for \emph{general deterministic} truthful auctions. We also show that the ratio of the learning rates of the bidders can \emph{qualitatively} affect the convergence of the bidders. Next, we consider the problem of revenue maximization in this environment. In the setting with fully rational bidders, \citet{myerson1981optimal} showed that revenue can be maximized by using a second-price auction with reserves.We show that, in stark contrast, in our setting with learning bidders, \emph{randomized} auctions can have strictly better revenue guarantees than second-price auctions with reserves, when $T$ is large enough. Finally, we study revenue maximization in the non-asymptotic regime. We define a notion of {\em auctioneer regret} comparing the revenue generated to the revenue of a second price auction with truthful bids. When the auctioneer has to use the same auction throughout the interaction, we show an (almost) tight regret bound of $\smash{\widetilde \Theta(T^{3/4})}.$ If the auctioneer can change auctions during the interaction, but in a way that is oblivious to the bids, we show an (almost) tight bound of $\smash{\widetilde \Theta(\sqrt{T})}.$

We measure the time-integrated CP asymmetry in $D^{0} \rightarrow K^{0}_{S}K^{0}_{S}$ decays reconstructed in $e^{+}e^{-} \rightarrow c\overline{c}$ events collected by the Belle and Belle II experiments. The corresponding data samples have integrated luminosities of 980 fb$^{-1}$ and 428 fb$^{-1}$, respectively. The $D^{0}$ decays are required to originate from the $D^{*+} \rightarrow D^{0}\pi^{+}$ decay, which determines the charm flavor at production time. A control sample of $D^{0} \rightarrow K^{+}K^{-}$ decays is used to correct for production and detection asymmetries. The result, $(-1.4\pm1.3{\rm(stat)}\pm0.1{\rm (syst)})\%$, is consistent with previous determinations and with CP symmetry., Comment: 10 pages, 3 figures. arXiv admin note: text overlap with arXiv:2410.22961

We perform a model-independent measurement of the $D^0$-$\overline{D}{}^0$ mixing parameters using samples of $e^+e^-$-collision data collected by the Belle and Belle II experiments that have integrated luminosities of $951\ \text{fb}^{-1}$ and $408\ \text{fb}^{-1}$, respectively. Approximately $2.05\times10^6$ neutral $D$ mesons are reconstructed in the $D^0\rightarrow K^0_{S}\pi^+\pi^-$ channel, with the neutral $D$ flavor tagged by the charge of the pion in the $D^{*+}\rightarrow D^0\pi^+$ decay. Assuming charge-parity symmetry, the mixing parameters are measured to be $ x = (4.0\pm1.7\pm0.4)\times 10^{-3} $ and $ y = (2.9\pm1.4\pm0.3)\times 10^{-3}$, where the first uncertainties are statistical and the second systematic. The results are consistent with previous determinations.

As the quality of large language models has improved, there has been increased interest in using them to model non-linguistic tokens. For example, the Decision Transformer recasts agentic decision making as a sequence modeling problem, using a decoder-only LLM to model the distribution over the discrete action space for an Atari agent. However, when adapting LLMs to non-linguistic domains, it remains unclear if softmax over discrete bins captures the continuous structure of the tokens and the potentially complex distributions needed for high quality token generation. We introduce a neural network layer, constructed using Fourier series, which we can easily substitute for any linear layer if we want the outputs to have a more continuous structure. We perform extensive analysis on synthetic datasets, as well as on large-scale decision making and time series forecasting tasks. We also provide theoretical evidence that this layer can better learn signal from data while ignoring high-frequency noise. All of our results support the effectiveness of our proposed Fourier head in scenarios where the underlying data distribution has a natural continuous structure. For example, the Fourier head improves a Decision Transformer agent's returns by 46% on the Atari Seaquest game, and increases a state-of-the-art times series foundation model's forecasting performance by 3.5% across 20 benchmarks unseen during training., Comment: Project page and code are at https://nategillman.com/fourier-head

In direct searches of dark matter, multi-scatter signatures are now being sought to probe scattering cross sections that are large enough to make the detector optically thick to incident particles. We provide some significant updates to the multi-scatter program. Using considerations of energy deposition, we derive the reaches in cross section and mass of various proposed large-volume detectors: a kilotonne fiducial mass "module of opportunity" at DUNE, a kilotonne xenon detector suggested for neutrinoless double beta decay, the gaseous detector CYGNUS, and the dark matter detectors XLZD and Argo. Where the velocity vector can be reconstructed event-by-event, the Galactic dark matter velocity distribution may be inferred. We exploit this to show that halo substructure such as tidal streams can be picked up if they make up about 10% of the local dark matter density., Comment: 9 pages revtex4 + references, 3 figures, 1 table

Retrieval-Augmented Generation (RAG) systems using large language models (LLMs) often generate inaccurate responses due to the retrieval of irrelevant or loosely related information. Existing methods, which operate at the document level, fail to effectively filter out such content. We propose LLM-driven chunk filtering, ChunkRAG, a framework that enhances RAG systems by evaluating and filtering retrieved information at the chunk level. Our approach employs semantic chunking to divide documents into coherent sections and utilizes LLM-based relevance scoring to assess each chunk's alignment with the user's query. By filtering out less pertinent chunks before the generation phase, we significantly reduce hallucinations and improve factual accuracy. Experiments show that our method outperforms existing RAG models, achieving higher accuracy on tasks requiring precise information retrieval. This advancement enhances the reliability of RAG systems, making them particularly beneficial for applications like fact-checking and multi-hop reasoning.

Preclinical diffusion MRI (dMRI) has proven value in methods development and validation, characterizing the biological basis of diffusion phenomena, and comparative anatomy. While dMRI enables in vivo non-invasive characterization of tissue, ex vivo dMRI is increasingly being used to probe tissue microstructure and brain connectivity. Ex vivo dMRI has several experimental advantages that facilitate high spatial resolution and high signal-to-noise ratio (SNR) images, cutting-edge diffusion contrasts, and direct comparison with histological data as a methodological validation. However, there are a number of considerations that must be made when performing ex vivo experiments. The steps from tissue preparation, image acquisition and processing, and interpretation of results are complex, with many decisions that not only differ dramatically from in vivo imaging of small animals, but ultimately affect what questions can be answered using the data. This work concludes a 3-part series of recommendations and considerations for preclinical dMRI. Herein, we describe best practices for dMRI of ex vivo tissue, with a focus on image pre-processing, data processing and model fitting, and tractography. In each section, we attempt to provide guidelines and recommendations, but also highlight areas for which no guidelines exist (and why), and where future work should lie. We end by providing guidelines on code sharing and data sharing, and point towards open-source software and databases specific to small animal and ex vivo imaging., Comment: Part 3 of 3 in "Considerations and recommendations for preclinical diffusion MRI"

Differentiating between Intestinal Tuberculosis (ITB) and Crohn's Disease (CD) poses a significant clinical challenge due to their similar symptoms, clinical presentations, and imaging features. This study leverages Computed Tomography Enterography (CTE) scans, deep learning, and traditional computer vision to address this diagnostic dilemma. A consensus among radiologists from renowned institutions has recognized the visceral-to-subcutaneous fat (VF/SF) ratio as a surrogate biomarker for differentiating between ITB and CD. Previously done manually, we propose a novel 2D image computer vision algorithm for auto-segmenting subcutaneous fat to automate this ratio calculation, enhancing diagnostic efficiency and objectivity. As a benchmark, we compare the results to those obtained using the TotalSegmentator tool, a popular deep learning-based software for automatic segmentation of anatomical structures, and manual calculations by radiologists. We also demonstrated the performance on 3D CT volumes using a slicing method and provided a benchmark comparison of the algorithm with the TotalSegmentator tool. Additionally, we propose a scoring approach to integrate scores from radiological features, such as the fat ratio and pulmonary TB probability, into a single score for diagnosis. We trained a ResNet10 model on a dataset of CTE scans with samples from ITB, CD, and normal patients, achieving an accuracy of 75%. To enhance interpretability and gain clinical trust, we integrated the explainable AI technique Grad-CAM with ResNet10 to explain the model's predictions. Due to the small dataset size (100 total cases), the feature-based scoring system is considered more reliable and trusted by radiologists compared to the deep learning model for disease diagnosis., Comment: 9 pages, 3 figures, 3 algorithms

Large Language Models (LLMs) demonstrate exceptional capabilities in a multitude of NLP tasks. However, the efficacy of such models to languages other than English is often limited. Prior works have shown that encoder-only models such as BERT or XLM-RoBERTa show impressive cross lingual transfer of their capabilities from English to other languages. In this work, we propose a pretraining strategy that uses active forgetting to achieve similar cross lingual transfer in decoder-only LLMs. We show that LLMs pretrained with active forgetting are highly effective when adapting to new and unseen languages. Through extensive experimentation, we find that LLMs pretrained with active forgetting are able to learn better multilingual representations which translates to better performance in many downstream tasks., Comment: 12 pages, 11 tables, 12 figures

A common challenge towards the adaptability of Large Language Models (LLMs) is their ability to learn new languages over time without hampering the model's performance on languages in which the model is already proficient (usually English). Continual fine-tuning (CFT) is the process of sequentially fine-tuning an LLM to enable the model to adapt to downstream tasks with varying data distributions and time shifts. This paper focuses on the language adaptability of LLMs through CFT. We study a two-phase CFT process in which an English-only end-to-end fine-tuned LLM from Phase 1 (predominantly Task Ability) is sequentially fine-tuned on a multilingual dataset -- comprising task data in new languages -- in Phase 2 (predominantly Language Ability). We observe that the ``similarity'' of Phase 2 tasks with Phase 1 determines the LLM's adaptability. For similar phase-wise datasets, the LLM after Phase 2 does not show deterioration in task ability. In contrast, when the phase-wise datasets are not similar, the LLM's task ability deteriorates. We test our hypothesis on the open-source \mis\ and \llm\ models with multiple phase-wise dataset pairs. To address the deterioration, we analyze tailored variants of two CFT methods: layer freezing and generative replay. Our findings demonstrate their effectiveness in enhancing the language ability of LLMs while preserving task performance, in comparison to relevant baselines., Comment: 19 pages, 6 tables, 4 figures

The importance of Reinforcement Learning from Human Feedback (RLHF) in aligning large language models (LLMs) with human values cannot be overstated. RLHF is a three-stage process that includes supervised fine-tuning (SFT), reward learning, and policy learning. Although there are several offline and online approaches to aligning LLMs, they often suffer from distribution shift issues. These issues arise from the inability to accurately capture the distributional interdependence between the reward learning and policy learning stages. Consequently, this has led to various approximated approaches, but the theoretical insights and motivations remain largely limited to tabular settings, which do not hold in practice. This gap between theoretical insights and practical implementations is critical. It is challenging to address this gap as it requires analyzing the performance of AI alignment algorithms in neural network-parameterized settings. Although bi-level formulations have shown promise in addressing distribution shift issues, they suffer from the hyper-gradient problem, and current approaches lack efficient algorithms to solve this. In this work, we tackle these challenges employing the bi-level formulation laid out in Kwon et al. (2024) along with the assumption \emph{Weak Gradient Domination} to demonstrate convergence in an RLHF setup, obtaining a sample complexity of $\epsilon^{-\frac{7}{2}}$ . Our key contributions are twofold: (i) We propose a bi-level formulation for AI alignment in parameterized settings and introduce a first-order approach to solve this problem. (ii) We analyze the theoretical convergence rates of the proposed algorithm and derive state-of-the-art bounds. To the best of our knowledge, this is the first work to establish convergence rate bounds and global optimality for the RLHF framework in neural network-parameterized settings.

Deriving inference from heterogeneous inputs (such as images, text, and audio) is an important skill for humans to perform day-to-day tasks. A similar ability is desirable for the development of advanced Artificial Intelligence (AI) systems. While state-of-the-art models are rapidly closing the gap with human-level performance on diverse computer vision and NLP tasks separately, they struggle to solve tasks that require joint reasoning over visual and textual modalities. Inspired by GLUE (Wang et. al., 2018)- a multitask benchmark for natural language understanding, we propose VL-GLUE in this paper. VL-GLUE consists of over 100k samples spanned across seven different tasks, which at their core require visuo-linguistic reasoning. Moreover, our benchmark comprises of diverse image types (from synthetically rendered figures, and day-to-day scenes to charts and complex diagrams) and includes a broad variety of domain-specific text (from cooking, politics, and sports to high-school curricula), demonstrating the need for multi-modal understanding in the real-world. We show that this benchmark is quite challenging for existing large-scale vision-language models and encourage development of systems that possess robust visuo-linguistic reasoning capabilities., Comment: 18 pages, 7 figures

We present a novel approach to automatically generate non-trivial task-specific synthetic datasets for hallucination detection. Our approach features a two-step generation-selection pipeline, using hallucination pattern guidance and a language style alignment during generation. Hallucination pattern guidance leverages the most important task-specific hallucination patterns while language style alignment aligns the style of the synthetic dataset with benchmark text. To obtain robust supervised detectors from synthetic datasets, we also adopt a data mixture strategy to improve performance robustness and generalization. Our results on three datasets show that our generated hallucination text is more closely aligned with non-hallucinated text versus baselines, to train hallucination detectors with better generalization. Our hallucination detectors trained on synthetic datasets outperform in-context-learning (ICL)-based detectors by a large margin of 32%. Our extensive experiments confirm the benefits of our approach with cross-task and cross-generator generalization. Our data-mixture-based training further improves the generalization and robustness of hallucination detection.

We use supersymmetry to enlarge the dimension of the Hilbert space on which the unitary evolution of the state of the quantum fields acts. We discuss how to control the unitary evolution or TPCP maps generated by the quantum evolution of the fields by controlling the vacuum expectations of other fields in the theory. This amounts to breaking supersymmetry using control vacuum expectation values of the other fields. The evolution of the wave functional or TPCP maps obtained by tracing out over other fields is based on the Feynman path integral formula for the fields. By using the methods of quantum stochastic filtering, we estimate the evolving state of the fields from non-demolition noise measurements and then design a family of TPCP maps evolving in time whose outputs match the estimated evolving state. In this way, we are able to simulate the evolution of the state of the quantum noisy fields. Direct matching of the designed TPCP map to output the evolving system state is not possible since there is no way by which we can determine the exact evolving state, we can only estimate it using non-demolition measurements. The family of designed TPCP maps can be based on using a simulated master equation with unknown parameters incorporated into the Hamiltonian and the other Lindblad operators, chosen so as to match the state outputted by the quantum filter.

Businesses can benefit from customer feedback in different modalities, such as text and images, to enhance their products and services. However, it is difficult to extract actionable and relevant pairs of text segments and images from customer feedback in a single pass. In this paper, we propose a novel multi-modal method that fuses image and text information in a latent space and decodes it to extract the relevant feedback segments using an image-text grounded text decoder. We also introduce a weakly-supervised data generation technique that produces training data for this task. We evaluate our model on unseen data and demonstrate that it can effectively mine actionable insights from multi-modal customer feedback, outperforming the existing baselines by $14$ points in F1 score., Comment: NAACL 2024

Summarizing customer feedback to provide actionable insights for products/services at scale is an important problem for businesses across industries. Lately, the review volumes are increasing across regions and languages, therefore the challenge of aggregating and understanding customer sentiment across multiple languages becomes increasingly vital. In this paper, we propose a novel framework involving a two-step paradigm \textit{Extract-then-Summarise}, namely MARS to revolutionise traditions and address the domain agnostic aspect-level multilingual review summarisation. Extensive automatic and human evaluation shows that our approach brings substantial improvements over abstractive baselines and efficiency to real-time systems., Comment: arXiv admin comment: This version has been removed by arXiv administrators as the submitter did not have the rights to agree to the license at the time of submission

We prove a sharp upper bound on the Hausdorff dimension of weighted singular vectors in $\mathbb{R}^m$ using dynamics on homogeneous spaces, specifically the method of integral inequalities. Together with the lower bound proved recently by Kim and Park \cite{KimPark2024}, this determines the Hausdorff dimension of weighted singular vectors, thereby generalizing to arbitrary dimension, the work of Liao, Shi, Solan, and Tamam \cite{LSST}, who determined the Hausdorff dimension of weighted singular vectors in two dimensions. We also provide the first known bounds for the Hausdorff dimension of weighted singular vectors restricted to fractal subsets., Comment: 19 Pages, Comments welcome

We present a measurement of the branching fraction and time-dependent charge-parity ($CP$) decay-rate asymmetries in $B^0 \to J/\psi \pi^0$ decays. The data sample was collected with the Belle~II detector at the SuperKEKB asymmetric $e^+e^-$ collider in 2019-2022 and contains $(387\pm 6)\times 10^6$ $B\overline{B}$ meson pairs from $\Upsilon(4S)$ decays. We reconstruct $392\pm 24$ signal decays and fit the $CP$ parameters from the distribution of the proper-decay-time difference of the two $B$ mesons. We measure the branching fraction to be $B(B^0 \to J/\psi \pi^0)=(2.02 \pm 0.12 \pm 0.10)\times 10^{-5}$ and the direct and mixing-induced $CP$ asymmetries to be $C_{CP}=0.13 \pm 0.12 \pm 0.03$ and $S_{CP}=-0.88 \pm 0.17 \pm 0.03$, respectively, where the first uncertainties are statistical and the second are systematic. We observe mixing-induced $CP$ violation with a significance of $5.0$ standard deviations for the first time in this mode.

In recommendation systems, there has been a growth in the number of recommendable items (# of movies, music, products). When the set of recommendable items is large, training and evaluation of item recommendation models becomes computationally expensive. To lower this cost, it has become common to sample negative items. However, the recommendation quality can suffer from biases introduced by traditional negative sampling mechanisms. In this work, we demonstrate the benefits from correcting the bias introduced by sampling of negatives. We first provide sampled batch version of the well-studied WARP and LambdaRank methods. Then, we present how these methods can benefit from improved ranking estimates. Finally, we evaluate the recommendation quality as a result of correcting rank estimates and demonstrate that WARP and LambdaRank can be learned efficiently with negative sampling and our proposed correction technique.

Knowledge graph (KG) completion aims to identify additional facts that can be inferred from the existing facts in the KG. Recent developments in this field have explored this task in the inductive setting, where at test time one sees entities that were not present during training; the most performant models in the inductive setting have employed path encoding modules in addition to standard subgraph encoding modules. This work similarly focuses on KG completion in the inductive setting, without the explicit use of path encodings, which can be time-consuming and introduces several hyperparameters that require costly hyperparameter optimization. Our approach uses a Transformer-based subgraph encoding module only; we introduce connection-biased attention and entity role embeddings into the subgraph encoding module to eliminate the need for an expensive and time-consuming path encoding module. Evaluations on standard inductive KG completion benchmark datasets demonstrate that our Connection-Biased Link Prediction (CBLiP) model has superior performance to models that do not use path information. Compared to models that utilize path information, CBLiP shows competitive or superior performance while being faster. Additionally, to show that the effectiveness of connection-biased attention and entity role embeddings also holds in the transductive setting, we compare CBLiP's performance on the relation prediction task in the transductive setting.

Retrieval Augmented Generation (RAG) is a widely used approach for leveraging external context in several natural language applications such as question answering and information retrieval. Yet, the exact nature in which a Language Model (LM) leverages this non-parametric memory or retrieved context isn't clearly understood. This paper mechanistically examines the RAG pipeline to highlight that LMs demonstrate a "shortcut'' effect and have a strong bias towards utilizing the retrieved context to answer questions, while relying minimally on model priors. We propose (a) Causal Mediation Analysis; for proving that parametric memory is minimally utilized when answering a question and (b) Attention Contributions and Knockouts for showing the last token residual stream do not get enriched from the subject token in the question, but gets enriched from tokens of RAG-context. We find this pronounced "shortcut'' behaviour to be true across both LLMs (e.g.,LlaMa) and SLMs (e.g., Phi), Comment: Accepted to Blackbox NLP @ EMNLP 2024

We consider the problem of a designer who wants to allocate resources to representatives, that then distribute the resources they receive among the individuals they represent. Motivated by the way Feeding America, one of the largest U.S. charities, allocates donations to food banks, which then further distribute the donations to food-insecure individuals, we focus on mechanisms that use artificial currencies. We compare auctions through the lens of the Price of Anarchy, with respect to three canonical welfare objectives: utilitarian social welfare (sum of individuals' utilities), Nash social welfare (product of individuals' utilities), and egalitarian social welfare (minimum of individuals' utilities). We prove strong lower bounds on the Price of Anarchy of all auctions that allocate each item to the highest bidder, subject to a mild technical constraint; this includes Feeding America's current auction, the First-Price auction. In sharp contrast, our main result shows that adapting the classic Trading Post mechanism of Shapley and Shubik to this setting, and coupled with Feeding America's choice of budget rule (each representative gets an amount of artificial currency equal to the number of individuals it represents), achieves a small Price of Anarchy for all generalized $p$-mean objectives simultaneously. Our bound on the Price of Anarchy of the Trading Post mechanism depends on $\ell$: the product of the rank and the ``incoherence'' of the underlying valuation matrix, which together capture a notion of how ``spread out'' the values of a matrix are. This notion has been extremely influential in the matrix completion literature, and, to the best of our knowledge, has never been used in auction theory prior to our work. Perhaps surprisingly, we prove that the dependence on $\ell$ is necessary: the Price of Anarchy of the Trading Post mechanism is $\Omega(\sqrt{\ell})$., Comment: The 20th Conference on Web and Internet Economics (WINE 2024)

The rapid progress of Large Language Models (LLMs) has opened up new opportunities across various domains and applications; yet it also presents challenges related to potential misuse. To mitigate such risks, red teaming has been employed as a proactive security measure to probe language models for harmful outputs via jailbreak attacks. However, current jailbreak attack approaches are single-turn with explicit malicious queries that do not fully capture the complexity of real-world interactions. In reality, users can engage in multi-turn interactions with LLM-based chat assistants, allowing them to conceal their true intentions in a more covert manner. To bridge this gap, we, first, propose a new jailbreak approach, RED QUEEN ATTACK. This method constructs a multi-turn scenario, concealing the malicious intent under the guise of preventing harm. We craft 40 scenarios that vary in turns and select 14 harmful categories to generate 56k multi-turn attack data points. We conduct comprehensive experiments on the RED QUEEN ATTACK with four representative LLM families of different sizes. Our experiments reveal that all LLMs are vulnerable to RED QUEEN ATTACK, reaching 87.62% attack success rate on GPT-4o and 75.4% on Llama3-70B. Further analysis reveals that larger models are more susceptible to the RED QUEEN ATTACK, with multi-turn structures and concealment strategies contributing to its success. To prioritize safety, we introduce a straightforward mitigation strategy called RED QUEEN GUARD, which aligns LLMs to effectively counter adversarial attacks. This approach reduces the attack success rate to below 1% while maintaining the model's performance across standard benchmarks. Full implementation and dataset are publicly accessible at https://github.com/kriti-hippo/red_queen.

By framing reinforcement learning as a sequence modeling problem, recent work has enabled the use of generative models, such as diffusion models, for planning. While these models are effective in predicting long-horizon state trajectories in deterministic environments, they face challenges in dynamic settings with moving obstacles. Effective collision avoidance demands continuous monitoring and adaptive decision-making. While replanning at every timestep could ensure safety, it introduces substantial computational overhead due to the repetitive prediction of overlapping state sequences -- a process that is particularly costly with diffusion models, known for their intensive iterative sampling procedure. We propose an adaptive generative planning approach that dynamically adjusts replanning frequency based on the uncertainty of action predictions. Our method minimizes the need for frequent, computationally expensive, and redundant replanning while maintaining robust collision avoidance performance. In experiments, we obtain a 13.5% increase in the mean trajectory length and a 12.7% increase in mean reward over long-horizon planning, indicating a reduction in collision rates and an improved ability to navigate the environment safely.

We perform the first search for $C\!P$ violation in ${D_{(s)}^{+}\to{}K_{S}^{0}K^{-}\pi^{+}\pi^{+}}$ decays. We use a combined data set from the Belle and Belle II experiments, which study $e^+e^-$ collisions at center-of-mass energies at or near the $\Upsilon(4S)$ resonance. We use 980 fb$^{-1}$ of data from Belle and 428 fb$^{-1}$ of data from Belle~II. We measure six $C\!P$-violating asymmetries that are based on triple products and quadruple products of the momenta of final-state particles, and also the particles' helicity angles. We obtain a precision at the level of 0.5% for $D^+\to{}K_{S}^{0}K^{-}\pi^{+}\pi^{+}$ decays, and better than 0.3% for $D^+_{s}\to{}K_{S}^{0}K^{-}\pi^{+}\pi^{+}$ decays. No evidence of $C\!P$ violation is found. Our results for the triple-product asymmetries are the most precise to date for singly-Cabibbo-suppressed $D^+$ decays. Our results for the other asymmetries are the first such measurements performed for charm decays., Comment: 21 pages, 10 figures

As LLMs evolve, significant effort is spent on manually crafting prompts. While existing prompt optimization methods automate this process, they rely solely on learning from incorrect samples, leading to a sub-optimal performance. Additionally, an unexplored challenge in the literature is prompts effective for prior models may not perform well on newer versions or different languages. We propose the Learning from Contrastive Prompts (LCP) framework to address these gaps, enhancing both prompt optimization and adaptation. LCP employs contrastive learning to generate effective prompts by analyzing patterns in good and bad prompt examples. Our evaluation on the Big-Bench Hard dataset shows that LCP has a win rate of over 76% over existing methods in prompt optimization and demonstrates strong adaptability across different model versions, families, and languages. LCP offers a systematic approach to prompt engineering, reducing manual effort in deploying LLMs across varied contexts.

This paper improves upon existing data pruning methods for image classification by introducing a novel pruning metric and pruning procedure based on importance sampling. The proposed pruning metric explicitly accounts for data separability, data integrity, and model uncertainty, while the sampling procedure is adaptive to the pruning ratio and considers both intra-class and inter-class separation to further enhance the effectiveness of pruning. Furthermore, the sampling method can readily be applied to other pruning metrics to improve their performance. Overall, the proposed approach scales well to high pruning ratio and generalizes better across different classification models, as demonstrated by experiments on four benchmark datasets, including the fine-grained classification scenario.

Existing subset selection methods for efficient learning predominantly employ discrete combinatorial and model-specific approaches which lack generalizability. For an unseen architecture, one cannot use the subset chosen for a different model. To tackle this problem, we propose $\texttt{SubSelNet}$, a trainable subset selection framework, that generalizes across architectures. Here, we first introduce an attention-based neural gadget that leverages the graph structure of architectures and acts as a surrogate to trained deep neural networks for quick model prediction. Then, we use these predictions to build subset samplers. This naturally provides us two variants of $\texttt{SubSelNet}$. The first variant is transductive (called as Transductive-$\texttt{SubSelNet}$) which computes the subset separately for each model by solving a small optimization problem. Such an optimization is still super fast, thanks to the replacement of explicit model training by the model approximator. The second variant is inductive (called as Inductive-$\texttt{SubSelNet}$) which computes the subset using a trained subset selector, without any optimization. Our experiments show that our model outperforms several methods across several real datasets, Comment: Published at NeurIPS 2023

Graph anomaly detection (GAD), which aims to identify unusual graph instances (nodes, edges, subgraphs, or graphs), has attracted increasing attention in recent years due to its significance in a wide range of applications. Deep learning approaches, graph neural networks (GNNs) in particular, have been emerging as a promising paradigm for GAD, owing to its strong capability in capturing complex structure and/or node attributes in graph data. Considering the large number of methods proposed for GNN-based GAD, it is of paramount importance to summarize the methodologies and findings in the existing GAD studies, so that we can pinpoint effective model designs for tackling open GAD problems. To this end, in this work we aim to present a comprehensive review of deep learning approaches for GAD. Existing GAD surveys are focused on task-specific discussions, making it difficult to understand the technical insights of existing methods and their limitations in addressing some unique challenges in GAD. To fill this gap, we first discuss the problem complexities and their resulting challenges in GAD, and then provide a systematic review of current deep GAD methods from three novel perspectives of methodology, including GNN backbone design, proxy task design for GAD, and graph anomaly measures. To deepen the discussions, we further propose a taxonomy of 13 fine-grained method categories under these three perspectives to provide more in-depth insights into the model designs and their capabilities. To facilitate the experiments and validation, we also summarize a collection of widely-used GAD datasets and empirical comparison. We further discuss multiple open problems to inspire more future high-quality research. A continuously updated repository for datasets, links to the codes of algorithms, and empirical comparison is available at https://github.com/mala-lab/Awesome-Deep-Graph-Anomaly-Detection., Comment: 24 pages, 6 figures, and 7 tables

Ranking a set of items based on their relevance to a given query is a core problem in search and recommendation. Transformer-based ranking models are the state-of-the-art approaches for such tasks, but they score each query-item independently, ignoring the joint context of other relevant items. This leads to sub-optimal ranking accuracy and high computational costs. In response, we propose Cross-encoders with Joint Efficient Modeling (CROSS-JEM), a novel ranking approach that enables transformer-based models to jointly score multiple items for a query, maximizing parameter utilization. CROSS-JEM leverages (a) redundancies and token overlaps to jointly score multiple items, that are typically short-text phrases arising in search and recommendations, and (b) a novel training objective that models ranking probabilities. CROSS-JEM achieves state-of-the-art accuracy and over 4x lower ranking latency over standard cross-encoders. Our contributions are threefold: (i) we highlight the gap between the ranking application's need for scoring thousands of items per query and the limited capabilities of current cross-encoders; (ii) we introduce CROSS-JEM for joint efficient scoring of multiple items per query; and (iii) we demonstrate state-of-the-art accuracy on standard public datasets and a proprietary dataset. CROSS-JEM opens up new directions for designing tailored early-attention-based ranking models that incorporate strict production constraints such as item multiplicity and latency.

In the rapidly evolving landscape of site reliability engineering (SRE), the demand for efficient and effective solutions to manage and resolve issues in site and cloud applications is paramount. This paper presents an innovative approach to action automation using large language models (LLMs) for script generation, assessment, and refinement. By leveraging the capabilities of LLMs, we aim to significantly reduce the human effort involved in writing and debugging scripts, thereby enhancing the productivity of SRE teams. Our experiments focus on Bash scripts, a commonly used tool in SRE, and involve the CodeSift dataset of 100 tasks and the InterCode dataset of 153 tasks. The results show that LLMs can automatically assess and refine scripts efficiently, reducing the need for script validation in an execution environment. Results demonstrate that the framework shows an overall improvement of 7-10% in script generation., Comment: Under Review

Motivated by online retail, we consider the problem of selling one item (e.g., an ad slot) to two non-excludable buyers (say, a merchant and a brand). This problem captures, for example, situations where a merchant and a brand cooperatively bid in an auction to advertise a product, and both benefit from the ad being shown. A mechanism collects bids from the two and decides whether to allocate and which payments the two parties should make. This gives rise to intricate incentive compatibility constraints, e.g., on how to split payments between the two parties. We approach the problem of finding a revenue-maximizing incentive-compatible mechanism from an online learning perspective; this poses significant technical challenges. First, the action space (the class of all possible mechanisms) is huge; second, the function that maps mechanisms to revenue is highly irregular, ruling out standard discretization-based approaches. In the stochastic setting, we design an efficient learning algorithm achieving a regret bound of $O(T^{3/4})$. Our approach is based on an adaptive discretization scheme of the space of mechanisms, as any non-adaptive discretization fails to achieve sublinear regret. In the adversarial setting, we exploit the non-Lipschitzness of the problem to prove a strong negative result, namely that no learning algorithm can achieve more than half of the revenue of the best fixed mechanism in hindsight. We then consider the $\sigma$-smooth adversary; we construct an efficient learning algorithm that achieves a regret bound of $O(T^{2/3})$ and builds on a succinct encoding of exponentially many experts. Finally, we prove that no learning algorithm can achieve less than $\Omega(\sqrt T)$ regret in both the stochastic and the smooth setting, thus narrowing the range where the minimax regret rates for these two problems lie., Comment: Paper accepted at ACM EC 2024

This application note discusses the working principle of spin-transfer torque magnetoresistive random access memory (STT-MRAM) and the impact that magnetic fields can have on STT-MRAM operation. Sources of magnetic field and typical magnitudes of magnetic fields are given. Based on the magnitude of commonly encountered external magnetic fields, we show below that magnetic immunity of STT-MRAM is sufficient for most uses once the chip is mounted on a printed circuit board (PCB) or inserted in its working environment. This statement is supported by the experience acquired during 60 years of use of magnetic hard disk drives (HDD) including 20 years of HDD with readers comprising magnetic tunnel junctions, 20+ years of use of magnetic field sensors as position encoders in automotive industry and 15+ years of use of MRAM. Mainly during chip handling does caution need to be exercised to avoid exposing the chip to excessively high magnetic fields.

Analysis of normalised C-moments of multiplicity distribution calculated from the different phenomenological models, the Bialas-Praszalowics~(BP) model, modified negative binomial and the superposed SGD at different center of mass~(cms) energies is presented.~The analysis covers a range of energies~(200-900 GeV) of $\bar{p}p$ collisions in restricted phase space slices.~A comparison of different models to the experimental data on charged particle multiplicity spectra from the $\bar{p}p$ annihilation in five pseudorapidity windows is reported.~The comparison shows that the two approaches other than the BP model are in better agreement to the data.~Results on variation of moments with pseudorapidity window size and with center of mass energy and observations from such a study in $\bar{p}p$ annihilation and $pp$ interactions at the same cms energy are also presented., Comment: 10 pages, 6 figures

Knowledge Organization Systems (KOSs), such as term lists, thesauri, taxonomies, and ontologies, play a fundamental role in categorising, managing, and retrieving information. In the academic domain, KOSs are often adopted for representing research areas and their relationships, primarily aiming to classify research articles, academic courses, patents, books, scientific venues, domain experts, grants, software, experiment materials, and several other relevant products and agents. These structured representations of research areas, widely embraced by many academic fields, have proven effective in empowering AI-based systems to i) enhance retrievability of relevant documents, ii) enable advanced analytic solutions to quantify the impact of academic research, and iii) analyse and forecast research dynamics. This paper aims to present a comprehensive survey of the current KOS for academic disciplines. We analysed and compared 45 KOSs according to five main dimensions: scope, structure, curation, usage, and links to other KOSs. Our results reveal a very heterogeneous scenario in terms of scope, scale, quality, and usage, highlighting the need for more integrated solutions for representing research knowledge across academic fields. We conclude by discussing the main challenges and the most promising future directions.

Ultralight scalar fields can experience runaway `superradiant' amplification near spinning black holes, resulting in a macroscopic `axion cloud' which slowly dissipates via continuous monochromatic gravitational waves. For a particular range of boson masses, $\mathcal{O}(10^{-11}$ -- $10^{-10})$ eV, an axion cloud will radiate in the $10$ -- $100$ kHz band of the Levitated Sensor Detector (LSD). Using fiducial models of the mass, spin, and age distributions of stellar-origin black holes, we simulate the present-day Milky Way population of these hypothetical objects. As a first step towards assessing the LSD's sensitivity to the resultant ensemble of GW signals, we compute the corresponding signal-to-noise ratios which build up over a nominal integration time of $10^{7}$ s, assuming the projected sensitivity of the $1$-m LSD prototype currently under construction, as well as for future $10$-m and $100$-m concepts. For a $100$-m cryogenic instrument, hundreds of resolvable signals could be expected if the boson mass $\mu$ is around $3\times10^{-11}$ eV, and this number diminishes with increasing $\mu$ up to $\approx 5.5\times10^{-11}$ eV. The much larger population of unresolved sources will produce a confusion foreground which could be detectable by a $10$-m instrument if $\mu \in (3-4.5)\times10^{-11}$ eV, or by a $100$-m instrument if $\mu \in (3-6)\times10^{-11}$ eV.

The structure of the low redshift Universe is dominated by a multi-scale void distribution delineated by filaments and walls of galaxies. The characteristics of voids; such as morphology, average density profile, and correlation function, can be used as cosmological probes. However, their physical properties are difficult to infer due to shot noise and the general lack of tracer particles used to define them. In this work, we construct a robust, topology-based void finding algorithm that utilizes Persistent Homology (PH) to detect persistent features in the data. We apply this approach to a volume limited sub-sample of galaxies in the SDSS I/II Main Galaxy catalog with the $r$-band absolute magnitude brighter than $M_r=-20.19$, and a set of mock catalogs constructed using the Horizon Run 4 cosmological $N$-body simulation. We measure the size distribution of voids, their averaged radial profile, sphericity, and the centroid nearest neighbor separation, using conservative values for the threshold and persistence. We find $32$ topologically robust voids in the SDSS data over the redshift range $0.02 \leq z \leq 0.116$, with effective radii in the range $21 - 56 \, h^{-1} \, {\rm Mpc}$. The median nearest neighbor void separation is found to be $\sim 57 \, h^{-1} \, {\rm Mpc}$, and the median radial void profile is consistent with the expected shape from the mock data., Comment: Accepted for publication in ApJ

The paper presents Maximal Ellipsoid Backward Reachable Trees MAXELLIPSOID BRT, which is a multi-query algorithm for planning of dynamic systems under stochastic motion uncertainty and constraints on the control input. In contrast to existing probabilistic planning methods that grow a roadmap of distributions, our proposed method introduces a framework to construct a roadmap of ambiguity sets of distributions such that each edge in our proposed roadmap provides a feasible control sequence for a family of distributions at once leading to efficient multi-query planning. Specifically, we construct a backward reachable tree of maximal size ambiguity sets and the corresponding distributionally robust edge controllers. Experiments show that the computation of these sets of distributions, in a backwards fashion from the goal, leads to efficient planning at a fraction of the size of the roadmap required for state-of-the-art methods. The computation of these maximal ambiguity sets and edges is carried out via a convex semidefinite relaxation to a novel nonlinear program. We also formally prove a theorem on maximum coverage for a technique proposed in our prior work., Comment: Submitted to IEEE Transactions on Automatic Control on June 26, 2024. arXiv admin note: substantial text overlap with arXiv:2403.14605