Search

We address the challenge of automated classification of diffuse large B-cell lymphoma (DLBCL) into its two primary subtypes: activated B-cell-like (ABC) and germinal center B-cell-like (GCB). Accurate classification between these subtypes is essential for determining the appropriate therapeutic strategy, given their distinct molecular profiles and treatment responses. Our proposed deep learning model demonstrates robust performance, achieving an average area under the curve (AUC) of (87.4 pm 5.7)\% during cross-validation. It shows a high positive predictive value (PPV), highlighting its potential for clinical application, such as triaging for molecular testing. To gain biological insights, we performed an analysis of morphological features of ABC and GCB subtypes. We segmented cell nuclei using a pre-trained deep neural network and compared the statistics of geometric and color features for ABC and GCB. We found that the distributions of these features were not very different for the two subtypes, which suggests that the visual differences between them are more subtle. These results underscore the potential of our method to assist in more precise subtype classification and can contribute to improved treatment management and outcomes for patients of DLBCL.

The Fermi-Hubbard model (FHM) is a simple yet rich model of strongly interacting electrons with complex dynamics and a variety of emerging quantum phases. These properties make it a compelling target for digital quantum simulation. Trotterization-based quantum simulations have shown promise, but implementations on current hardware are limited by noise, necessitating error mitigation techniques like circuit optimization and post-selection. A mapping of the FHM to a Z2 LGT was recently proposed that restricts the dynamics to a subspace protected by additional symmetries, and its ability for post-selection error mitigation was verified through noisy classical simulations. In this work, we propose and demonstrate a suite of algorithm-hardware co-design strategies on a trapped-ion quantum computer, targeting two key aspects of NISQ-era quantum simulation: circuit compilation and error mitigation. In particular, a novel combination of iteratively preconditioned gradient descent (IPG) and subsystem von Neumann Entropy compression reduces the 2-qubit gate count of FHM quantum simulation by 35%, consequently doubling the number of simulatable Trotter steps when used in tandem with error mitigation based on conserved symmetries, debiasing and sharpening techniques. Our work demonstrates the value of algorithm-hardware co-design to operate digital quantum simulators at the threshold of maximum circuit depths allowed by current hardware, and is broadly generalizable to strongly correlated systems in quantum chemistry and materials science.

Understanding animal behaviour is central to predicting, understanding, and mitigating impacts of natural and anthropogenic changes on animal populations and ecosystems. However, the challenges of acquiring and processing long-term, ecologically relevant data in wild settings have constrained the scope of behavioural research. The increasing availability of Unmanned Aerial Vehicles (UAVs), coupled with advances in machine learning, has opened new opportunities for wildlife monitoring using aerial tracking. However, limited availability of datasets with wild animals in natural habitats has hindered progress in automated computer vision solutions for long-term animal tracking. Here we introduce BuckTales, the first large-scale UAV dataset designed to solve multi-object tracking (MOT) and re-identification (Re-ID) problem in wild animals, specifically the mating behaviour (or lekking) of blackbuck antelopes. Collected in collaboration with biologists, the MOT dataset includes over 1.2 million annotations including 680 tracks across 12 high-resolution (5.4K) videos, each averaging 66 seconds and featuring 30 to 130 individuals. The Re-ID dataset includes 730 individuals captured with two UAVs simultaneously. The dataset is designed to drive scalable, long-term animal behaviour tracking using multiple camera sensors. By providing baseline performance with two detectors, and benchmarking several state-of-the-art tracking methods, our dataset reflects the real-world challenges of tracking wild animals in socially and ecologically relevant contexts. In making these data widely available, we hope to catalyze progress in MOT and Re-ID for wild animals, fostering insights into animal behaviour, conservation efforts, and ecosystem dynamics through automated, long-term monitoring., Comment: 9 pages, 5 figures

The operation efficiency of the electric transportation, energy storage, and grids mainly depends on the fundamental characteristics of the employed batteries. Fundamental variables like voltage, current, temperature, and estimated parameters, like the State of Charge (SoC) of the battery pack, influence the functionality of the system. This motivates the implementation of a Battery Management System (BMS), critical for managing and maintaining the health, safety, and performance of a battery pack. This is ensured by measuring parameters like temperature, cell voltage, and pack current. It also involves monitoring insulation levels and fire hazards, while assessing the prevailing useful life of the batteries and estimating the SoC and State of Health (SoH). Additionally, the system manages and controls key activities like cell balancing and charge/discharge processes. Thus functioning of the battery can be optimised, by guaranteeing the vital parameters to be well within the prescribed range. This article discusses the several cell balancing schemes, and focuses on the intricacies of cell balancing algorithms and optimisation methods for cell balancing. We begin surveying recent cell balancing algorithms and then provide selection guidelines taking into account their advantages, disadvantages, and applications. Finally, we discuss various optimization algorithms and outline the essential parameters involved in the cell balancing process., Comment: 33 pages, 8 figures, 14 tables, and 13 equations

Recent advancements in radiance field rendering show promising results in 3D scene representation, where Gaussian splatting-based techniques emerge as state-of-the-art due to their quality and efficiency. Gaussian splatting is widely used for various applications, including 3D human representation. However, previous 3D Gaussian splatting methods either use parametric body models as additional information or fail to provide any underlying structure, like human biomechanical features, which are essential for different applications. In this paper, we present a novel approach called HFGaussian that can estimate novel views and human features, such as the 3D skeleton, 3D key points, and dense pose, from sparse input images in real time at 25 FPS. The proposed method leverages generalizable Gaussian splatting technique to represent the human subject and its associated features, enabling efficient and generalizable reconstruction. By incorporating a pose regression network and the feature splatting technique with Gaussian splatting, HFGaussian demonstrates improved capabilities over existing 3D human methods, showcasing the potential of 3D human representations with integrated biomechanics. We thoroughly evaluate our HFGaussian method against the latest state-of-the-art techniques in human Gaussian splatting and pose estimation, demonstrating its real-time, state-of-the-art performance.

Topological effects manifest in a variety of lattice geometries. While square lattices, due to their simplicity, have been used for models supporting nontrivial topology, several exotic topological phenomena such as Dirac points, Weyl points and Haldane phases are most commonly supported by non-square lattices. Examples of prototypical non-square lattices include the honeycomb lattice of graphene and the Kagome lattice, both of which break fundamental symmetries and can exhibit quantized transport, especially when long-range hoppings and gauge fields are incorporated. The challenge of controllably realizing long-range hoppings and gauge fields has motivated a large body of research focused on harnessing lattices encoded in "synthetic" dimensions. Photons in particular have many internal degrees of freedom and hence show promise for implementing these synthetic dimensions; however, most photonic synthetic dimensions has hitherto created 1D or 2D square lattices. Here we show that non-square lattice Hamiltonians can be implemented using Floquet synthetic dimensions. Our construction uses dynamically modulated ring resonators and provides the capacity for direct $k$-space engineering of lattice Hamiltonians. Such a construction lifts constraints on the orthogonality of lattice vectors that make square geometries simpler to implement, and instead transfers the complexity to the engineering of complex Floquet drive signals. We simulate topological signatures of the Haldane and the brick-wall Haldane model and observe them to be robust in the presence of external optical drive and photon loss, and discuss unique characteristics of their topological transport when implemented on these Floquet lattices. Our proposal demonstrates the potential of driven-dissipative Floquet synthetic dimensions as a new architecture for $k$-space Hamiltonian simulation of high-dimensional lattice geometries.

A spectre is haunting consensus protocols-the spectre of adversary majority. The literature is inconclusive, with possibilities and impossibilities running abound. Dolev and Strong in 1983 showed an early possibility for up to 99% adversaries. Yet, we have known impossibility results for adversaries above 1/2 in synchrony, and above 1/3 in partial synchrony. What gives? It is high time that we pinpoint the culprit of this confusion: the critical role of the modeling details of clients. Are the clients sleepy or always-on? Are they silent or communicating? Can validators be sleepy too? We systematize models for consensus across four dimensions (sleepy/always-on clients, silent/communicating clients, sleepy/always-on validators, and synchrony/partial-synchrony), some of which are new, and tightly characterize the achievable safety and liveness resiliences with matching possibilities and impossibilities for each of the sixteen models. To this end, we unify folklore and earlier results, and fill gaps left in the literature with new protocols and impossibility theorems.

Ateniese, Goodrich, Lekakis, Papamanthou, Paraskevas, and Tamassia introduced the Accountable Storage protocol, which is a way for a client to outsource their data to a cloud storage provider while allowing the client to periodically perform accountability challenges. An accountability challenge efficiently recovers any pieces of data the server has lost or corrupted, allowing the client to extract the original copies of the damaged or lost data objects. A severe limitation of the prior accountable storage scheme of Ateniese et al., however, is that it is not fully dynamic. That is, it does not allow a client to freely insert and delete data from the outsourced data set after initializing the protocol, giving the protocol limited practical use in the real world. In this paper, we present Dynamic Accountable Storage, which is an efficient way for a client to periodically audit their cloud storage while also supporting insert and delete operations on the data set. To do so, we introduce a data structure, the IBLT tree, which allows either the server or the client to reconstruct data the server has lost or corrupted in a space-efficient way., Comment: 20 pages, 1 figure, AlgoCloud 2024

Recent advances in generative modeling with diffusion processes (DPs) enabled breakthroughs in image synthesis. Despite impressive image quality, these models have various prompt compliance problems, including low recall in generating multiple objects, difficulty in generating text in images, and meeting constraints like object locations and pose. For fine-grained editing and manipulation, they also require fine-grained semantic or instance maps that are tedious to produce manually. While prompt compliance can be enhanced by addition of loss functions at inference, this is time consuming and does not scale to complex scenes. To overcome these limitations, this work introduces a new family of \textit{Factor Graph Diffusion Models} (FG-DMs) that models the joint distribution of images and conditioning variables, such as semantic, sketch, depth or normal maps via a factor graph decomposition. This joint structure has several advantages, including support for efficient sampling based prompt compliance schemes, which produce images of high object recall, semi-automated fine-grained editing, text-based editing of conditions with noise inversion, explainability at intermediate levels, ability to produce labeled datasets for the training of downstream models such as segmentation or depth, training with missing data, and continual learning where new conditioning variables can be added with minimal or no modifications to the existing structure. We propose an implementation of FG-DMs by adapting a pre-trained Stable Diffusion (SD) model to implement all FG-DM factors, using only COCO dataset, and show that it is effective in generating images with 15\% higher recall than SD while retaining its generalization ability. We introduce an attention distillation loss that encourages consistency among the attention maps of all factors, improving the fidelity of the generated conditions and image., Comment: Accepted to NeurIPS 2024 conference. Project Page: https://deepaksridhar.github.io/factorgraphdiffusion.github.io/

Given two sets of elements (such as cell types and drug compounds), researchers typically only have access to a limited subset of their interactions. The task of causal imputation involves using this subset to predict unobserved interactions. Squires et al. (2022) have proposed two estimators for this task based on the synthetic interventions (SI) estimator: SI-A (for actions) and SI-C (for contexts). We extend their work and introduce a novel causal imputation estimator, generalized synthetic interventions (GSI). We prove the identifiability of this estimator for data generated from a more complex latent factor model. On synthetic and real data we show empirically that it recovers or outperforms their estimators.

Dual-arm manipulation is an area of growing interest in the robotics community. Enabling robots to perform tasks that require the coordinated use of two arms, is essential for complex manipulation tasks such as handling large objects, assembling components, and performing human-like interactions. However, achieving effective dual-arm manipulation is challenging due to the need for precise coordination, dynamic adaptability, and the ability to manage interaction forces between the arms and the objects being manipulated. We propose a novel pipeline that combines the advantages of policy learning based on environment feedback and gradient-based optimization to learn controller gains required for the control outputs. This allows the robotic system to dynamically modulate its impedance in response to task demands, ensuring stability and dexterity in dual-arm operations. We evaluate our pipeline on a trajectory-tracking task involving a variety of large, complex objects with different masses and geometries. The performance is then compared to three other established methods for controlling dual-arm robots, demonstrating superior results.

The accurate prediction of danger levels in video content is critical for enhancing safety and security systems, particularly in environments where quick and reliable assessments are essential. In this study, we perform a comparative analysis of various machine learning and deep learning models to predict danger ratings in a custom dataset of 100 videos, each containing 50 frames, annotated with human-rated danger scores ranging from 0 to 10. The danger ratings are further classified into three categories: no alert (less than 7)and high alert (greater than equal to 7). Our evaluation covers classical machine learning models, such as Support Vector Machines, as well as Neural Networks, and transformer-based models. Model performance is assessed using standard metrics such as accuracy, F1-score, and mean absolute error (MAE), and the results are compared to identify the most robust approach. This research contributes to developing a more accurate and generalizable danger assessment framework for video-based risk detection.

This paper introduces MotionGlot, a model that can generate motion across multiple embodiments with different action dimensions, such as quadruped robots and human bodies. By leveraging the well-established training procedures commonly used in large language models (LLMs), we introduce an instruction-tuning template specifically designed for motion-related tasks. Our approach demonstrates that the principles underlying LLM training can be successfully adapted to learn a wide range of motion generation tasks across multiple embodiments with different action dimensions. We demonstrate the various abilities of MotionGlot on a set of 6 tasks and report an average improvement of 35.3% across tasks. Additionally, we contribute two new datasets: (1) a dataset of expert-controlled quadruped locomotion with approximately 48,000 trajectories paired with direction-based text annotations, and (2) a dataset of over 23,000 situational text prompts for human motion generation tasks. Finally, we conduct hardware experiments to validate the capabilities of our system in real-world applications.

Distinguishing between swarming and swimming, the two principal forms of bacterial movement, holds significant conceptual and clinical relevance. This is because bacteria that exhibit swarming capabilities often possess unique properties crucial to the pathogenesis of infectious diseases and may also have therapeutic potential. Here, we report a deep learning-based swarming classifier that rapidly and autonomously predicts swarming probability using a single blurry image. Compared with traditional video-based, manually-processed approaches, our method is particularly suited for high-throughput environments and provides objective, quantitative assessments of swarming probability. The swarming classifier demonstrated in our work was trained on Enterobacter sp. SM3 and showed good performance when blindly tested on new swarming (positive) and swimming (negative) test images of SM3, achieving a sensitivity of 97.44% and a specificity of 100%. Furthermore, this classifier demonstrated robust external generalization capabilities when applied to unseen bacterial species, such as Serratia marcescens DB10 and Citrobacter koseri H6. It blindly achieved a sensitivity of 97.92% and a specificity of 96.77% for DB10, and a sensitivity of 100% and a specificity of 97.22% for H6. This competitive performance indicates the potential to adapt our approach for diagnostic applications through portable devices or even smartphones. This adaptation would facilitate rapid, objective, on-site screening for bacterial swarming motility, potentially enhancing the early detection and treatment assessment of various diseases, including inflammatory bowel diseases (IBD) and urinary tract infections (UTI)., Comment: 17 Pages, 4 Figures

Receptor occupancy (RO) studies using PET neuroimaging play a critical role in the development of drugs targeting the central nervous system (CNS). The conventional approach to estimate drug receptor occupancy consists in estimation of binding potential changes between two PET scans (baseline and post-drug injection). This estimation is typically performed separately for each scan by first reconstructing dynamic PET scan data before fitting a kinetic model to time activity curves. This approach fails to properly model the noise in PET measurements, resulting in poor RO estimates, especially in low receptor density regions. Objective: In this work, we evaluate a novel joint direct parametric reconstruction framework to directly estimate distributions of RO and other kinetic parameters in the brain from a pair of baseline and post-drug injection dynamic PET scans. Methods: The proposed method combines the use of regularization on RO maps with alternating optimization to enable estimation of occupancy even in low binding regions. Results: Simulation results demonstrate the quantitative improvement of this method over conventional approaches in terms of accuracy and precision of occupancy. The proposed method is also evaluated in preclinical in-vivo experiments using 11C-MK-6884 and a muscarinic acetylcholine receptor 4 positive allosteric modulator drug, showing improved estimation of receptor occupancy as compared to traditional estimators. Conclusion: The proposed joint direct estimation framework improves RO estimation compared to conventional methods, especially in intermediate to low-binding regions. Significance: This work could potentially facilitate the evaluation of new drug candidates targeting the CNS.

We propose a zero-shot text-driven 3D shape deformation system that deforms an input 3D mesh of a manufactured object to fit an input text description. To do this, our system optimizes the parameters of a deformation model to maximize an objective function based on the widely used pre-trained vision language model CLIP. We find that CLIP-based objective functions exhibit many spurious local optima; to circumvent them, we parameterize deformations using a novel deformation model called BoxDefGraph which our system automatically computes from an input mesh, the BoxDefGraph is designed to capture the object aligned rectangular/circular geometry features of most manufactured objects. We then use the CMA-ES global optimization algorithm to maximize our objective, which we find to work better than popular gradient-based optimizers. We demonstrate that our approach produces appealing results and outperforms several baselines.

We introduce Controller-Embedded Language Model Interactions (CELI), a framework that integrates control logic directly within language model (LM) prompts, facilitating complex, multi-stage task execution. CELI addresses limitations of existing prompt engineering and workflow optimization techniques by embedding control logic directly within the operational context of language models, enabling dynamic adaptation to evolving task requirements. Our framework transfers control from the traditional programming execution environment to the LMs, allowing them to autonomously manage computational workflows while maintaining seamless interaction with external systems and functions. CELI supports arbitrary function calls with variable arguments, bridging the gap between LMs' adaptive reasoning capabilities and conventional software paradigms' structured control mechanisms. To evaluate CELI's versatility and effectiveness, we conducted case studies in two distinct domains: code generation (HumanEval benchmark) and multi-stage content generation (Wikipedia-style articles). The results demonstrate notable performance improvements across a range of domains. CELI achieved a 4.9 percentage point improvement over the best reported score of the baseline GPT-4 model on the HumanEval code generation benchmark. In multi-stage content generation, 94.4% of CELI-produced Wikipedia-style articles met or exceeded first draft quality when optimally configured, with 44.4% achieving high quality. These outcomes underscore CELI's potential for optimizing AI-driven workflows across diverse computational domains., Comment: 26 pages, 2 figures

The discrete empirical interpolation method (DEIM) is a well-established approach, widely used for state reconstruction using sparse sensor/measurement data, nonlinear model reduction, and interpretable feature selection. We introduce the tensor t-product Q-DEIM (t-Q-DEIM), an extension of the DEIM framework for dealing with tensor-valued data. The proposed approach seeks to overcome one of the key drawbacks of DEIM, viz., the need for matricizing the data, which can distort any structural and/or geometric information. Our method leverages the recently developed tensor t-product algebra to avoid reshaping the data. In analogy with the standard DEIM, we formulate and solve a tensor-valued least-squares problem, whose solution is achieved through an interpolatory projection. We develop a rigorous, computable upper bound for the error resulting from the t-Q-DEIM approximation. Using five different tensor-valued datasets, we numerically illustrate the better approximation properties of t-Q-DEIM and the significant computational cost reduction it offers., Comment: 37 pages, 22 figures, 1 table

The goal of machine learning is generalization. While the No Free Lunch Theorem states that we cannot obtain theoretical guarantees for generalization without further assumptions, in practice we observe that simple models which explain the training data generalize best: a principle called Occam's razor. Despite the need for simple models, most current approaches in machine learning only minimize the training error, and at best indirectly promote simplicity through regularization or architecture design. Here, we draw a connection between Occam's razor and in-context learning: an emergent ability of certain sequence models like Transformers to learn at inference time from past observations in a sequence. In particular, we show that the next-token prediction loss used to train in-context learners is directly equivalent to a data compression technique called prequential coding, and that minimizing this loss amounts to jointly minimizing both the training error and the complexity of the model that was implicitly learned from context. Our theory and the empirical experiments we use to support it not only provide a normative account of in-context learning, but also elucidate the shortcomings of current in-context learning methods, suggesting ways in which they can be improved. We make our code available at https://github.com/3rdCore/PrequentialCode.

Triangular cross-section nanodevices are among the leading approaches for integrating color centers with photonics for applications in quantum information processing. We design periodic and aperiodic fishbone triangular grating couplers in silicon carbide. We optimize the designs for achieving up to $\sim$ 31% collection efficiency from color center integrated triangular devices to a microscopy system. Using an ion beam angle etching process, we demonstrate proof-of-principle fabrication of the designed devices for future implementation in wafer-scale quantum nanophotonics.

Large Language Models (LLMs) have fast become an essential tools to many conversational chatbots due to their ability to provide coherent answers for varied queries. Datasets used to train these LLMs are often a mix of generic and synthetic samples, thus lacking the verification needed to provide correct and verifiable answers for T.V. News. We collect and share a large collection of QA pairs extracted from transcripts of news recordings from various news-channels across the United States. Resultant QA pairs are then used to fine-tune an off-the-shelf LLM model. Our model surpasses base models of similar size on several open LLM benchmarks. We further integrate and propose a RAG method to improve contextualization of our answers and also point it to a verifiable news recording., Comment: 5 pages, under review at ICASSP 2025

Scientific research often seeks to understand the causal structure underlying high-level variables in a system. For example, climate scientists study how phenomena, such as El Ni\~no, affect other climate processes at remote locations across the globe. However, scientists typically collect low-level measurements, such as geographically distributed temperature readings. From these, one needs to learn both a mapping to causally-relevant latent variables, such as a high-level representation of the El Ni\~no phenomenon and other processes, as well as the causal model over them. The challenge is that this task, called causal representation learning, is highly underdetermined from observational data alone, requiring other constraints during learning to resolve the indeterminacies. In this work, we consider a temporal model with a sparsity assumption, namely single-parent decoding: each observed low-level variable is only affected by a single latent variable. Such an assumption is reasonable in many scientific applications that require finding groups of low-level variables, such as extracting regions from geographically gridded measurement data in climate research or capturing brain regions from neural activity data. We demonstrate the identifiability of the resulting model and propose a differentiable method, Causal Discovery with Single-parent Decoding (CDSD), that simultaneously learns the underlying latents and a causal graph over them. We assess the validity of our theoretical results using simulated data and showcase the practical validity of our method in an application to real-world data from the climate science field., Comment: 33 pages, 17 figures

The Small and Large Magellanic Clouds are the only galaxies outside our own in which radio pulsars have been discovered to date. The sensitivity of the MeerKAT radio interferometer offers an opportunity to search for a population of more distant extragalactic pulsars. The TRAPUM (TRansients And PUlsars with MeerKAT) collaboration has performed a radio-domain search for pulsars and transients in the dwarf star-forming galaxies Sextans A and B, situated at the edge of the local group 1.4 Mpc away. We conducted three 2-hour multi-beam observations at L-band (856-1712 MHz) with the full array of MeerKAT. No pulsars were found down to a radio pseudo-luminosity upper limit of 7.9$\pm$0.4 Jy kpc$^{2}$ at 1400 MHz, which is 28 times more sensitive than the previous limit from the Murriyang telescope. This luminosity is 30 per cent greater than that of the brightest known radio pulsar and sets a cut-off on the luminosity distributions of the entire Sextans A and B galaxies for unobscured radio pulsars beamed in our direction. A Fast Radio Burst was detected in one of the Sextans A observations at a Dispersion Measure (DM) of 737 pc cm$^{-3}$. We believe this is a background event not associated with the dwarf galaxy due to its large DM and its S/N being strongest in the wide-field incoherent beam of MeerKAT., Comment: 11 pages, 9 figures, 5 tables. Accepted for publication in Monthly Notices of the Royal Astronomical Society

We present a novel 2D convex hull peeling algorithm for outlier detection, which repeatedly removes the point on the hull that decreases the hull's area the most. To find k outliers among n points, one simply peels k points. The algorithm is an efficient heuristic for exact methods, which find the k points whose removal together results in the smallest convex hull. Our algorithm runs in O(nlogn) time using O(n) space for any choice of k. This is a significant speedup compared to the fastest exact algorithms, which run in O(n^2logn + (n - k)^3) time using O(n\logn + (n-k)^3) space by Eppstein et al., and O(nlogn + 4k_C_2k (3k)^k n) time by Atanassov et al. Existing heuristic peeling approaches are not area-based. Instead, an approach by Harsh et al. repeatedly removes the point furthest from the mean using various distance metrics and runs in O(nlogn + kn) time. Other approaches greedily peel one convex layer at a time, which is efficient when using an O(nlogn) time algorithm by Chazelle to compute the convex layers. However, in many cases this fails to recover outliers. For most values of n and k, our approach is the fastest and first practical choice for finding outliers based on minimizing the area of the convex hull. Our algorithm also generalizes to other objectives such as perimeter., Comment: 8 pages. 5 figures. Published at the 2024 Canadian Conference of Computational Geometry

The world is filled with a wide variety of objects. For robots to be useful, they need the ability to find arbitrary objects described by people. In this paper, we present LeLaN(Learning Language-conditioned Navigation policy), a novel approach that consumes unlabeled, action-free egocentric data to learn scalable, language-conditioned object navigation. Our framework, LeLaN leverages the semantic knowledge of large vision-language models, as well as robotic foundation models, to label in-the-wild data from a variety of indoor and outdoor environments. We label over 130 hours of data collected in real-world indoor and outdoor environments, including robot observations, YouTube video tours, and human walking data. Extensive experiments with over 1000 real-world trials show that our approach enables training a policy from unlabeled action-free videos that outperforms state-of-the-art robot navigation methods, while being capable of inference at 4 times their speed on edge compute. We open-source our models, datasets and provide supplementary videos on our project page (https://learning-language-navigation.github.io/)., Comment: 23 pages, 9 figures, 5 tables, Conference on Robot Learning 2024

We generalize Yekutieli-Zhang's noncommutative Serre Duality Theorem to the setting of noncommutative spaces associated to dg-algebras. As an application, we establish some finiteness properties of derived global sections over such noncommutative spaces. Along the way, we generalize Yekutieli's notion of a balanced dualizing complex to the setting of dg-algebras and establish some cases in which they exist., Comment: 14 pages

Diffusion models have recently surpassed GANs in image synthesis and editing, offering superior image quality and diversity. However, achieving precise control over attributes in generated images remains a challenge. Concept Sliders introduced a method for fine-grained image control and editing by learning concepts (attributes/objects). However, this approach adds parameters and increases inference time due to the loading and unloading of Low-Rank Adapters (LoRAs) used for learning concepts. These adapters are model-specific and require retraining for different architectures, such as Stable Diffusion (SD) v1.5 and SD-XL. In this paper, we propose a straightforward textual inversion method to learn concepts through text embeddings, which are generalizable across models that share the same text encoder, including different versions of the SD model. We refer to our method as Prompt Sliders. Besides learning new concepts, we also show that Prompt Sliders can be used to erase undesirable concepts such as artistic styles or mature content. Our method is 30% faster than using LoRAs because it eliminates the need to load and unload adapters and introduces no additional parameters aside from the target concept text embedding. Each concept embedding only requires 3KB of storage compared to the 8922KB or more required for each LoRA adapter, making our approach more computationally efficient. Project Page: https://deepaksridhar.github.io/promptsliders.github.io/, Comment: ECCV'24 - Unlearning and Model Editing Workshop. Code: https://github.com/DeepakSridhar/promptsliders

LLMs can now act as autonomous agents that interact with digital environments and complete specific objectives (e.g., arranging an online meeting). However, accuracy is still far from satisfactory, partly due to a lack of large-scale, direct demonstrations for digital tasks. Obtaining supervised data from humans is costly, and automatic data collection through exploration or reinforcement learning relies on complex environmental and content setup, resulting in datasets that lack comprehensive coverage of various scenarios. On the other hand, there is abundant knowledge that may indirectly assist task completion, such as online tutorials that were created for human consumption. In this work, we present Synatra, an approach that effectively transforms this indirect knowledge into direct supervision at scale. We define different types of indirect knowledge, and carefully study the available sources to obtain it, methods to encode the structure of direct demonstrations, and finally methods to transform indirect knowledge into direct demonstrations. We use 100k such synthetically-created demonstrations to finetune a 7B CodeLlama, and demonstrate that the resulting agent surpasses all comparably sized models on three web-based task benchmarks Mind2Web, MiniWoB++ and WebArena, as well as surpassing GPT-3.5 on WebArena and Mind2Web. In addition, while synthetic demonstrations prove to be only 3% the cost of human demonstrations (at $0.031 each), we show that the synthetic demonstrations can be more effective than an identical number of human demonstrations collected from limited domains.

Given a smooth proper morphism $f\colon X\rightarrow S$, we introduce a certain derived category where morphisms are permitted to be $\mathcal{O}_S$-linear differential operators. We then prove a generalisation of Serre duality that applies to two-term complexes of this type. We apply this to give a new proof of Poincar\'e duality for relative algebraic de Rham cohomology., Comment: 26 pages

Large language models (LLMs) are increasingly employed in information-seeking and decision-making tasks. Despite their broad utility, LLMs tend to generate information that conflicts with real-world facts, and their persuasive style can make these inaccuracies appear confident and convincing. As a result, end-users struggle to consistently align the confidence expressed by LLMs with the accuracy of their predictions, often leading to either blind trust in all outputs or a complete disregard for their reliability. In this work, we explore supervised finetuning on uncertainty-augmented predictions as a method to develop models that produce linguistic expressions of uncertainty. Specifically, we measure the calibration of pre-trained models and then fine-tune language models to generate calibrated linguistic expressions of uncertainty. Through experiments on various question-answering datasets, we demonstrate that LLMs are well-calibrated in assessing their predictions, and supervised finetuning based on the model's own confidence leads to well-calibrated expressions of uncertainty, particularly for single-claim answers.

The identification of persistent radio sources (PRSs) coincident with two repeating fast radio bursts (FRBs) supports FRB theories requiring a compact central engine. However, deep non-detections in other cases highlight the diversity of repeating FRBs and their local environments. Here, we perform a systematic search for radio sources towards 37 CHIME/FRB repeaters using their arcminute localizations and a combination of archival surveys and targeted observations. Through multi-wavelength analysis of individual radio sources, we identify two (20181030A-S1 and 20190417A-S1) for which we disfavor an origin of either star formation or an active galactic nucleus in their host galaxies and thus consider them candidate PRSs. We do not find any associated PRSs for the majority of the repeating FRBs in our sample. For 8 FRB fields with Very Large Array imaging, we provide deep limits on the presence of PRSs that are 2--4 orders of magnitude fainter than the PRS associated with FRB\,20121102A. Using Very Large Array Sky Survey imaging of all 37 fields, we constrain the rate of luminous ($\gtrsim$10$^{40}$ erg s$^{-1}$) PRSs associated with repeating FRBs to be low. Within the context of FRB-PRS models, we find that 20181030A-S1 and 20190417A-S1 can be reasonably explained within the context of magnetar, hypernebulae, gamma-ray burst afterglow, or supernova ejecta models -- although we note that both sources follow the radio luminosity versus rotation measure relationship predicted in the nebula model framework. Future observations will be required to both further characterize and confirm the association of these PRS candidates with the FRBs., Comment: 37 pages, 12 figures

Human auditory perception is compositional in nature -- we identify auditory streams from auditory scenes with multiple sound events. However, such auditory scenes are typically represented using clip-level representations that do not disentangle the constituent sound sources. In this work, we learn source-centric audio representations where each sound source is represented using a distinct, disentangled source embedding in the audio representation. We propose two novel approaches to learning source-centric audio representations: a supervised model guided by classification and an unsupervised model guided by feature reconstruction, both of which outperform the baselines. We thoroughly evaluate the design choices of both approaches using an audio classification task. We find that supervision is beneficial to learn source-centric representations, and that reconstructing audio features is more useful than reconstructing spectrograms to learn unsupervised source-centric representations. Leveraging source-centric models can help unlock the potential of greater interpretability and more flexible decoding in machine listening., Comment: Submitted to ICASSP 2025

We develop conservative tests for the mean of a bounded population using data from a stratified sample. The sample may be drawn sequentially, with or without replacement. The tests are "anytime valid," allowing optional stopping and continuation in each stratum. We call this combination of properties sequential, finite-sample, nonparametric validity. The methods express a hypothesis about the population mean as a union of intersection hypotheses describing within-stratum means. They test each intersection hypothesis using independent test supermartingales (TSMs) combined across strata by multiplication. The $P$-value of the global null hypothesis is then the maximum $P$-value of any intersection hypothesis in the union. This approach has three primary moving parts: (i) the rule for deciding which stratum to draw from next to test each intersection null, given the sample so far; (ii) the form of the TSM for each null in each stratum; and (iii) the method of combining evidence across strata. These choices interact. We examine the performance of a variety of rules with differing computational complexity. Approximately optimal methods have a prohibitive computational cost, while naive rules may be inconsistent -- they will never reject for some alternative populations, no matter how large the sample. We present a method that is statistically comparable to optimal methods in examples where optimal methods are computable, but computationally tractable for arbitrarily many strata. In numerical examples its expected sample size is substantially smaller than that of previous methods., Comment: 43 pages, 8 figures

The Audio Question Answering task includes audio event classification, audio captioning, and open ended reasoning. Recently, Audio Question Answering has garnered attention due to the advent of Large Audio Language Models. Current literature focuses on constructing LALMs by integrating audio encoders with text only Large Language Models through a projection module. While Large Audio Language Models excel in general audio understanding, they are limited in temporal reasoning which may hinder their commercial applications and on device deployment. This paper addresses these challenges and limitations in audio temporal reasoning. First, we introduce a data augmentation technique for generating reliable audio temporal questions and answers using an LLM. Second, we propose a continued finetuning curriculum learning strategy to specialize in temporal reasoning without compromising performance on finetuned tasks. Finally, we develop a reliable and transparent automated metric, assisted by an LLM, to measure the correlation between Large Audio Language Model responses and ground truth data intelligently. We demonstrate the effectiveness of our proposed techniques using SOTA LALMs on public audio benchmark datasets., Comment: 5 pages, 3 figures

Insects excel in trajectory and attitude handling during flight, yet the specific kinematic behaviours they use for maintaining stability in air disturbances are not fully understood. This study investigates the adaptive strategies of bumblebees when exposed to gust disturbances directed from three different angles within a plane cross-sectional to their flight path. By analyzing characteristic wing motions during gust traversal, we aim to uncover the mechanisms that enable bumblebees to maintain control in unsteady environments. We utilised high-speed cameras to capture detailed flight paths, allowing us to extract dynamic information. Our results reveal that bees make differential bilateral kinematic adjustments based on gust direction: sideward gusts elicit posterior shifts in the wing closest to the gust, while upward gusts trigger coordinated posterior shifts in both wings. Downward gusts prompted broader flapping and increased flapping frequencies, along with variations in flap timing and sweep angle. Stroke sweep angle was a primary factor influencing recovery responses, coupled with motion around the flap axis. The adaptive behaviours strategically position the wings to optimize gust reception and enhance wing-generated forces. These strategies can be distilled into specific behavioural patterns for analytical modelling to inform the design of robotic flyers. We observed a characteristic posterior shift of wings when particular counteractive manoeuvres were required. This adjustment reduced the portion of the stroke during which the wing receiving gust forces was positioned in front of the centre of gravity, potentially enhancing manoeuvrability and enabling more effective recovery manoeuvres. These findings deepen our understanding of insect flight dynamics and offer promising strategies for enhancing the stability and manoeuvrability of MAVs in turbulent environments., Comment: 8 pages, 6 figures