Search

This paper introduces AI Guide Dog (AIGD), a lightweight egocentric navigation assistance system for visually impaired individuals, designed for real-time deployment on smartphones. AIGD addresses key challenges in blind navigation by employing a vision-only, multi-label classification approach to predict directional commands, ensuring safe traversal across diverse environments. We propose a novel technique to enable goal-based outdoor navigation by integrating GPS signals and high-level directions, while also addressing uncertain multi-path predictions for destination-free indoor navigation. Our generalized model is the first navigation assistance system to handle both goal-oriented and exploratory navigation scenarios across indoor and outdoor settings, establishing a new state-of-the-art in blind navigation. We present methods, datasets, evaluations, and deployment insights to encourage further innovations in assistive navigation systems.

Integrating multiple observational studies for meta-analysis has sparked much interest. The presented R package WMAP (Weighted Meta-Analysis with Pseudo-Population) addresses a critical gap in the implementation of integrative weighting approaches for multiple observational studies and causal inferences about various groups of subjects, such as disease subtypes. The package features three weighting approaches, each representing a special case of the unified weighting framework introduced by Guha and Li (2024), which includes an extension of inverse probability weights for data integration settings. It performs meta-analysis on user-inputted datasets as follows: (i) it first estimates the propensity scores for study-group combinations, calculates subject balancing weights, and determines the effective sample size (ESS) for a user-specified weighting method; and (ii) it then estimates various features of multiple counterfactual group outcomes, such as group medians and differences in group means for the mRNA expression of eight genes. Additionally, bootstrap variability estimates are provided. Among the implemented weighting methods, we highlight the FLEXible, Optimized, and Realistic (FLEXOR) method, which is specifically designed to maximize the ESS within the unified framework. The use of the software is illustrated by simulations as well as a multi-site breast cancer study conducted in seven medical centers.

The gut microbiome significantly influences responses to cancer therapies, including immunotherapies, primarily through its impact on the metabolome. Despite some existing studies addressing the effects of specific microbial genera on individual metabolites, there is little to no prior work focused on identifying the key microbiome components at the genus level that shape the overall metabolome profile. To bridge this gap, we introduce B-MASTER (Bayesian Multivariate regression Analysis for Selecting Targeted Essential Regressors), a fully Bayesian framework incorporating an L1 penalty to promote sparsity in the coefficient matrix and an L2 penalty to shrink coefficients for non-major covariate components simultaneously, thereby isolating essential regressors. The method is complemented with a scalable Gibbs sampling algorithm, whose computational speed increases linearly with the number of parameters and remains largely unaffected by sample size and data-specific characteristics for models of fixed dimensions. Notably, B-MASTER achieves full posterior inference for models with up to four million parameters within a practical time-frame. Using this approach, we identify key microbial genera influencing the overall metabolite profile, conduct an in-depth analysis of their effects on the most abundant metabolites, and investigate metabolites differentially abundant in colorectal cancer patients. These results provide foundational insights into the impact of the microbiome at the genus level on metabolite profiles relevant to cancer, a relationship that remains largely unexplored in the existing literature.

A Markov model is a widely used tool for modeling sequences of events from a finite state-space and hence can be employed to identify the transition probabilities across treatments based on treatment sequence data. To understand how patient-level covariates impact these treatment transitions, the transition probabilities are modeled as a function of patient covariates. This approach enables the visualization of the effect of patient-level covariates on the treatment transitions across patient visits. The proposed method automatically estimates the entries of the transition matrix with smaller numbers of empirical transitions as constant; the user can set desired cutoff of the number of empirical transition counts required for a particular transition probability to be estimated as a function of covariates. Firstly, this strategy automatically enforces the final estimated transition matrix to contain zeros at the locations corresponding to zero empirical transition counts, avoiding further complicated model constructs to handle sparsity, in an efficient manner. Secondly, it restricts estimation of transition probabilities as a function of covariates, when the number of empirical transitions is particularly small, thus avoiding the identifiability issue which might arise due to the p>n scenario when estimating each transition probability as a function of patient covariates. To optimize the multi-modal likelihood, a parallelized scalable global optimization routine is also developed. The proposed method is applied to understand how the transitions across disease modifying therapies (DMTs) in Multiple Sclerosis (MS) patients are influenced by patient-level demographic and clinical phenotypes.

We present a benchmark for Planning And Reasoning Tasks in humaN-Robot collaboration (PARTNR) designed to study human-robot coordination in household activities. PARTNR tasks exhibit characteristics of everyday tasks, such as spatial, temporal, and heterogeneous agent capability constraints. We employ a semi-automated task generation pipeline using Large Language Models (LLMs), incorporating simulation in the loop for grounding and verification. PARTNR stands as the largest benchmark of its kind, comprising 100,000 natural language tasks, spanning 60 houses and 5,819 unique objects. We analyze state-of-the-art LLMs on PARTNR tasks, across the axes of planning, perception and skill execution. The analysis reveals significant limitations in SoTA models, such as poor coordination and failures in task tracking and recovery from errors. When LLMs are paired with real humans, they require 1.5x as many steps as two humans collaborating and 1.1x more steps than a single human, underscoring the potential for improvement in these models. We further show that fine-tuning smaller LLMs with planning data can achieve performance on par with models 9 times larger, while being 8.6x faster at inference. Overall, PARTNR highlights significant challenges facing collaborative embodied agents and aims to drive research in this direction., Comment: Alphabetical author order

In retail sales forecasting, accurately predicting future sales is crucial for inventory management and strategic planning. Traditional methods like LR often fall short due to the complexity of sales data, which includes seasonality and numerous product families. Recent advancements in machine learning (ML) provide more robust alternatives. This research benefits from the power of ML, particularly Random Forest (RF), Gradient Boosting (GB), Support Vector Regression (SVR), and XGBoost, to improve prediction accuracy. Despite advancements, a significant gap exists in handling complex datasets with high seasonality and multiple product families. The proposed solution involves implementing and optimizing a RF model, leveraging hyperparameter tuning through randomized search cross-validation. This approach addresses the complexities of the dataset, capturing intricate patterns that traditional methods miss. The optimized RF model achieved an R-squared value of 0.945, substantially higher than the initial RF model and traditional LR, which had an R-squared of 0.531. The model reduced the root mean squared logarithmic error (RMSLE) to 1.172, demonstrating its superior predictive capability. The optimized RF model did better than cutting-edge models like Gradient Boosting (R-squared: 0.942), SVR (R-squared: 0.940), and XGBoost (R-squared: 0.939), with more minor mean squared error (MSE) and mean absolute error (MAE) numbers. The results demonstrate that the optimized RF model excels in forecasting retail sales, handling the datasets complexity with higher accuracy and reliability. This research highlights the importance of advanced ML techniques in predictive analytics, offering a significant improvement over traditional methods and other contemporary models., Comment: 2024 4th International Conference on Sustainable Expert Systems (ICSES)

Most recommender systems research focuses on binary historical user-item interaction encodings to predict future interactions. User features, item features, and interaction strengths remain largely under-utilized in this space or only indirectly utilized, despite proving largely effective in large-scale production recommendation systems. We propose a new attention mechanism, loosely based on the principles of collaborative filtering, called Row-Column Separable Attention RCSA to take advantage of real-valued interaction weights as well as user and item features directly. Building on this mechanism, we additionally propose a novel Graph Diffusion Transformer GDiT architecture which is trained to iteratively denoise the weighted interaction matrix of the user-item interaction graph directly. The weighted interaction matrix is built from the bipartite structure of the user-item interaction graph and corresponding edge weights derived from user-item rating interactions. Inspired by the recent progress in text-conditioned image generation, our method directly produces user-item rating predictions on the same scale as the original ratings by conditioning the denoising process on user and item features with a principled approach., Comment: 6 pages, 13 figures

In this work, we study for the first time jet-medium interactions in heavy-ion collisions with introduction of saturation and Sudakov effects with parameters tuned for upcoming forward calorimeter acceptances in experiments, in particular the ALICE FoCal detector. We focus on $\gamma +$jet correlations by taking into account in-medium parton evolution using the BDIM equation that describes jet interactions with the quark-gluon plasma (QGP) combined with vacuum-like emissions (VLE). We systematically introduce the early time gluon saturation dynamics through the small-$x$ Improved Transverse Momentum Dependent factorization (ITMD). For our purpose, we use Monte Carlo programs KATIE and TMDICE to generate hard events and in-medium parton evolution, respectively. We present results of azimuthal correlations and nuclear modification ratios to gauge the impact of the gluon saturation effects at early time for the in-medium jet energy loss., Comment: 13 pages, 6 figures, missing references added

In heavy-ion collisions, nuclear matter is subjected to extreme conditions in a highly dynamical, rapidly evolving environment. This poses a tremendous challenge for calculating jet quenching observables. Current approaches rely on analytical results for static cases, introducing theoretical uncertainties and biases in our understanding of the pre-equilibrated medium. To address this issue, we employ resummation schemes to derive analytical rates for radiative energy loss in generic, evolving backgrounds. We investigate regimes where rare scattering and multiple scattering with the dynamical medium occurs, and extract relevant scales governing the in-medium emission rate of soft gluons. Our analysis indicates that strong jet quenching is only possible when the equilibration time of the medium is longer than its mean free path, highlighting the importance of medium modifications of jets in the earliest stages of heavy-ion collisions. We also demonstrate analytically that a medium evolution, which initially has a small coupling to jets, typically leads to a stronger jet azimuthal asymmetry at the same jet suppression factor., Comment: 6 pages, 3 figures and 2 supplemental material pages

The integration of machine learning (ML) models enhances the efficiency, affordability, and reliability of feature detection in microscopy, yet their development and applicability are hindered by the dependency on scarce and often flawed manually labeled datasets and a lack of domain awareness. We addressed these challenges by creating a physics-based synthetic image and data generator, resulting in a machine learning model that achieves comparable precision (0.86), recall (0.63), F1 scores (0.71), and engineering property predictions (R2=0.82) to a model trained on human-labeled data. We enhanced both models by using feature prediction confidence scores to derive an image-wide confidence metric, enabling simple thresholding to eliminate ambiguous and out-of-domain images resulting in performance boosts of 5-30% with a filtering-out rate of 25%. Our study demonstrates that synthetic data can eliminate human reliance in ML and provides a means for domain awareness in cases where many feature detections per image are needed.

Language is never spoken in a vacuum. It is expressed, comprehended, and contextualized within the holistic backdrop of the speaker's history, actions, and environment. Since humans are used to communicating efficiently with situated language, the practicality of robotic assistants hinge on their ability to understand and act upon implicit and situated instructions. In traditional instruction following paradigms, the agent acts alone in an empty house, leading to language use that is both simplified and artificially "complete." In contrast, we propose situated instruction following, which embraces the inherent underspecification and ambiguity of real-world communication with the physical presence of a human speaker. The meaning of situated instructions naturally unfold through the past actions and the expected future behaviors of the human involved. Specifically, within our settings we have instructions that (1) are ambiguously specified, (2) have temporally evolving intent, (3) can be interpreted more precisely with the agent's dynamic actions. Our experiments indicate that state-of-the-art Embodied Instruction Following (EIF) models lack holistic understanding of situated human intention., Comment: European Conference on Computer Vision 2024 (ECCV 2024)

This article focuses on training work carried out in artificial intelligence (AI) at the National Center for Supercomputing Applications (NCSA) at the University of Illinois Urbana-Champaign via a research experience for undergraduates (REU) program named FoDOMMaT. It also describes why we are interested in AI, and concludes by discussing what we've learned from running this program and its predecessor over six years.

The presented paper introduces a design for a phase-locked loop (PLL) that is utilized in frequency synthesis and modulation-demodulation within communication systems and in VLSI applications. The CMOS PLL is designed using 180 nm Fabrication Technology on Cadence Virtuoso Tool with a supply voltage of 1.8 V. The performance is evaluated through simulations and measurements, which demonstrate its ability to track and lock onto the input frequency. The PLL is a frequency synthesizer implemented to generate 2.4 GHz frequency. The input reference clock from a crystal oscillator is 150 MHz square wave. Negative feedback is given by divide-by-16 frequency divider, ensuring the phase and frequency synchronization between the divided signal and the reference signal. The design has essential components such as a phase frequency detector, charge pump, loop filter, current-starved voltage-controlled oscillator (CSVCO), and frequency divider. Through their collaborative operation, the system generates an output frequency that is 16 times the input frequency. The centre frequency of the 3-stage CSVCO is 3.208 GHz at 900 mV input voltage. With an input voltage ranging from 0.4 V to 1.8 V, the VCO offers a tuning range that spans from 1.066 GHz to 3.731 GHz.PLL demonstrates a lock-in range spanning from 70.4 MHz to 173 MHz, with an output frequency range of 1.12 GHz to 2.78 GHz. It achieves a lock time of 260.03 ns and consumes a maximum power of 5.15 mW at 2.4 GHz.

The need for abundant labelled data in supervised Adversarial Training (AT) has prompted the use of Self-Supervised Learning (SSL) techniques with AT. However, the direct application of existing SSL methods to adversarial training has been sub-optimal due to the increased training complexity of combining SSL with AT. A recent approach, DeACL, mitigates this by utilizing supervision from a standard SSL teacher in a distillation setting, to mimic supervised AT. However, we find that there is still a large performance gap when compared to supervised adversarial training, specifically on larger models. In this work, investigate the key reason for this gap and propose Projected Feature Adversarial Training (ProFeAT) to bridge the same. We show that the sub-optimal distillation performance is a result of mismatch in training objectives of the teacher and student, and propose to use a projection head at the student, that allows it to leverage weak supervision from the teacher while also being able to learn adversarially robust representations that are distinct from the teacher. We further propose appropriate attack and defense losses at the feature and projector, alongside a combination of weak and strong augmentations for the teacher and student respectively, to improve the training data diversity without increasing the training complexity. Through extensive experiments on several benchmark datasets and models, we demonstrate significant improvements in both clean and robust accuracy when compared to existing SSL-AT methods, setting a new state-of-the-art. We further report on-par/ improved performance when compared to TRADES, a popular supervised-AT method.

Despite the considerable advancements in English LLMs, the progress in building comparable models for other languages has been hindered due to the scarcity of tailored resources. Our work aims to bridge this divide by introducing an expansive suite of resources specifically designed for the development of Indic LLMs, covering 22 languages, containing a total of 251B tokens and 74.8M instruction-response pairs. Recognizing the importance of both data quality and quantity, our approach combines highly curated manually verified data, unverified yet valuable data, and synthetic data. We build a clean, open-source pipeline for curating pre-training data from diverse sources, including websites, PDFs, and videos, incorporating best practices for crawling, cleaning, flagging, and deduplication. For instruction-fine tuning, we amalgamate existing Indic datasets, translate/transliterate English datasets into Indian languages, and utilize LLaMa2 and Mixtral models to create conversations grounded in articles from Indian Wikipedia and Wikihow. Additionally, we address toxicity alignment by generating toxic prompts for multiple scenarios and then generate non-toxic responses by feeding these toxic prompts to an aligned LLaMa2 model. We hope that the datasets, tools, and resources released as a part of this work will not only propel the research and development of Indic LLMs but also establish an open-source blueprint for extending such efforts to other languages. The data and other artifacts created as part of this work are released with permissive licenses., Comment: ACL-2024 Outstanding Paper

The availability of Large Language Models (LLMs) which can generate code, has made it possible to create tools that improve developer productivity. Integrated development environments or IDEs which developers use to write software are often used as an interface to interact with LLMs. Although many such tools have been released, almost all of them focus on general-purpose programming languages. Domain-specific languages, such as those crucial for Information Technology (IT) automation, have not received much attention. Ansible is one such YAML-based IT automation-specific language. Ansible Lightspeed is an LLM-based service designed explicitly to generate Ansible YAML, given natural language prompt. In this paper, we present the design and implementation of the Ansible Lightspeed service. We then evaluate its utility to developers using diverse indicators, including extended utilization, analysis of user edited suggestions, as well as user sentiments analysis. The evaluation is based on data collected for 10,696 real users including 3,910 returning users. The code for Ansible Lightspeed service and the analysis framework is made available for others to use. To our knowledge, our study is the first to involve thousands of users of code assistants for domain-specific languages. We are also the first code completion tool to present N-Day user retention figures, which is 13.66% on Day 30. We propose an improved version of user acceptance rate, called Strong Acceptance rate, where a suggestion is considered accepted only if less than 50% of it is edited and these edits do not change critical parts of the suggestion. By focusing on Ansible, Lightspeed is able to achieve a strong acceptance rate of 49.08% for multi-line Ansible task suggestions. With our findings we provide insights into the effectiveness of small, dedicated models in a domain-specific context., Comment: This paper has been published at the 39th IEEE/ACM International Conference on Automated Software Engineering (ASE 2024), Industry Showcase under the title "Ansible Lightspeed: A Code Generation Service for IT Automation"

In this paper, we consider the two-sample location shift model, a classic semiparametric model introduced by Stein (1956). This model is known for its adaptive nature, enabling nonparametric estimation with full parametric efficiency. Existing nonparametric estimators of the location shift often depend on external tuning parameters, which restricts their practical applicability (Van der Vaart and Wellner, 2021). We demonstrate that introducing an additional assumption of log-concavity on the underlying density can alleviate the need for tuning parameters. We propose a one step estimator for location shift estimation, utilizing log-concave density estimation techniques to facilitate tuning-free estimation of the efficient influence function. While we employ a truncated version of the one step estimator for theoretical adaptivity, our simulations indicate that the one step estimators perform best with zero truncation, eliminating the need for tuning during practical implementation., Comment: Corrected minor algebraic mistakes in the proof of Theorem A.1

Research on data generation and augmentation has been focused majorly on enhancing generation models, leaving a notable gap in the exploration and refinement of methods for evaluating synthetic data. There are several text similarity metrics within the context of generated data filtering which can impact the performance of specific Natural Language Understanding (NLU) tasks, specifically focusing on intent and sentiment classification. In this study, we propose RankAug, a text-ranking approach that detects and filters out the top augmented texts in terms of being most similar in meaning with lexical and syntactical diversity. Through experiments conducted on multiple datasets, we demonstrate that the judicious selection of filtering techniques can yield a substantial improvement of up to 35% in classification accuracy for under-represented classes., Comment: Accepted at the GEM workshop at EMNLP 2023

The Koopman operator presents an attractive approach to achieve global linearization of nonlinear systems, making it a valuable method for simplifying the understanding of complex dynamics. While data-driven methodologies have exhibited promise in approximating finite Koopman operators, they grapple with various challenges, such as the judicious selection of observables, dimensionality reduction, and the ability to predict complex system behaviors accurately. This study presents a novel approach termed Mori-Zwanzig autoencoder (MZ-AE) to robustly approximate the Koopman operator in low-dimensional spaces. The proposed method leverages a nonlinear autoencoder to extract key observables for approximating a finite invariant Koopman subspace and integrates a non-Markovian correction mechanism using the Mori-Zwanzig formalism. Consequently, this approach yields a closed representation of dynamics within the latent manifold of the nonlinear autoencoder, thereby enhancing the precision and stability of the Koopman operator approximation. Demonstrations showcase the technique's ability to capture regime transitions in the flow around a cylinder. It also provides a low dimensional approximation for Kuramoto-Sivashinsky with promising short-term predictability and robust long-term statistical performance. By bridging the gap between data-driven techniques and the mathematical foundations of Koopman theory, MZ-AE offers a promising avenue for improved understanding and prediction of complex nonlinear dynamics., Comment: 22 pages, 11 figures

Millisecond pulsars (MSP) are an important subclass of rotation powered pulsars (RPP), traditionally defined as those with $P_s < 20-30$~ms and $B_s \lesssim 10^{10}$~G. We re-examine this definition by applying Gaussian mixture model (GMM) analysis to identify distinct clusters within the RPP population and find that the MSPs appear to be better demarcated by the condition $\mathbf{P_s \lesssim 16}$~ms., Comment: 3 pages, 1 figure

The lack of quality labeled data is one of the main bottlenecks for training Deep Learning models. As the task increases in complexity, there is a higher penalty for overfitting and unstable learning. The typical paradigm employed today is Self-Supervised learning, where the model attempts to learn from a large corpus of unstructured and unlabeled data and then transfer that knowledge to the required task. Some notable examples of self-supervision in other modalities are BERT for Large Language Models, Wav2Vec for Speech Recognition, and the Masked AutoEncoder for Vision, which all utilize Transformers to solve a masked prediction task. GeoAI is uniquely poised to take advantage of the self-supervised methodology due to the decades of data collected, little of which is precisely and dependably annotated. Our goal is to extract building and road segmentations from Digital Elevation Models (DEM) that provide a detailed topography of the earths surface. The proposed architecture is the Masked Autoencoder pre-trained on ImageNet (with the limitation that there is a large domain discrepancy between ImageNet and DEM) with an UperNet Head for decoding segmentations. We tested this model with 450 and 50 training images only, utilizing roughly 5% and 0.5% of the original data respectively. On the building segmentation task, this model obtains an 82.1% Intersection over Union (IoU) with 450 Images and 69.1% IoU with only 50 images. On the more challenging road detection task the model obtains an 82.7% IoU with 450 images and 73.2% IoU with only 50 images. Any hand-labeled dataset made today about the earths surface will be immediately obsolete due to the constantly changing nature of the landscape. This motivates the clear necessity for data-efficient learners that can be used for a wide variety of downstream tasks.

Sessile hydrogel drops on rigid surfaces exhibit wetting/contact morphology intermediate between liquid drops and glass spheres. Using density functional theory, we reveal the contact forces acting between a hydrogel and a rigid glass surface. We show that while transitioning from liquid-like to solid-like hydrogels, there exists a critical hydrogel elasticity which enables a switch from attractive to repulsive interaction with the underlying rigid glass surface. Our theoretical model is validated by experimental observations of sessile Polyacrylamide (PAAm) hydrogels of varying elasticity on glass surfaces. Further, the proposed model successfully approaches Young's law in the pure liquid limit and work of adhesion in the glassy limit. Lastly, we show a modified contact angle relation taking into account the hydrogel elasticity to explain the features of a distinct hydrogel foot.

This open-access dataset provides consistent records of air travel volumes between 205 airport catchments in Colombia and the associated number of reported human cases of Zika virus within these catchments during the arbovirus outbreak between October 2015 and September 2016. We associated in this dataset the monthly air travel volumes provided by the Colombian Civil Aviation Authority (AEROCIVIL) with the reported human cases of Zika Virus published by The Pan American Health Organization (PAHO). Our methodology consists of geocoding all the reported airports and identifying the catchment of each airport using the municipalities' boundaries since reported human cases of Zika Virus are available at the municipal level. In addition, we calculated the total population at risk in each airport catchment by combining the total population count in a catchment with the environmental suitability of the Aedes aegypti mosquito, the vector for the Zika virus. We separated the monthly air travel volumes into domestic and international based on the location of the origin airport. The current dataset includes the total air travel volumes of 23,539,364 passengers on domestic flights and 11,592,197 on international ones. We validated our dataset by comparing the monthly aggregated air travel volumes between airport catchments to those predicted by the gravity model. We hope the novel dataset will provide a resource to researchers studying the role of human mobility in the spread of mosquito-borne diseases and modeling disease spread in realistic networks.

Variational Quantum algorithms, especially Quantum Approximate Optimization and Variational Quantum Eigensolver (VQE) have established their potential to provide computational advantage in the realm of combinatorial optimization. However, these algorithms suffer from classically intractable gradients limiting the scalability. This work addresses the scalability challenge for VQE by proposing a classical gradient computation method which utilizes the parameter shift rule but computes the expected values from the circuits using a tensor ring approximation. The parametrized gates from the circuit transform the tensor ring by contracting the matrix along the free edges of the tensor ring. While the single qubit gates do not alter the ring structure, the state transformations from the two qubit rotations are evaluated by truncating the singular values thereby preserving the structure of the tensor ring and reducing the computational complexity. This variation of the Matrix product state approximation grows linearly in number of qubits and the number of two qubit gates as opposed to the exponential growth in the classical simulations, allowing for a faster evaluation of the gradients on classical simulators., Comment: 12 pages, 13 figures, preprint

Surface codes are one of the most important topological stabilizer codes in the theory of quantum error correction. In this paper, we provide an efficient way to obtain surface codes through Measurement-based quantum computation (MBQC) using cluster state as the resource state. Simple twodimensional surface codes are studied and analyzed using stabilizer formalism. We also present an algorithm to computationally obtain the stabilizer of the surface codes, through which we later determine the distance of the codes. We note the difference in the stabilizers of the surface codes obtained by Fowler et al. wherein they used CNOT entangling operation to create the resource state as opposed to the cluster state which is formed using CZ entangling operation. We provide a theoretical calculation to understand this difference. The obtained surface codes can be used practically as an encoder circuit to encode one logical qubit., Comment: 33 pages, 14 figures

Despite great strides in language-guided manipulation, existing work has been constrained to table-top settings. Table-tops allow for perfect and consistent camera angles, properties are that do not hold in mobile manipulation. Task plans that involve moving around the environment must be robust to egocentric views and changes in the plane and angle of grasp. A further challenge is ensuring this is all true while still being able to learn skills efficiently from limited data. We propose Spatial-Language Attention Policies (SLAP) as a solution. SLAP uses three-dimensional tokens as the input representation to train a single multi-task, language-conditioned action prediction policy. Our method shows an 80% success rate in the real world across eight tasks with a single model, and a 47.5% success rate when unseen clutter and unseen object configurations are introduced, even with only a handful of examples per task. This represents an improvement of 30% over prior work (20% given unseen distractors and configurations). We see a 4x improvement over baseline in mobile manipulation setting. In addition, we show how SLAPs robustness allows us to execute Task Plans from open-vocabulary instructions using a large language model for multi-step mobile manipulation. For videos, see the website: https://robotslap.github.io

We examine electron-transport coefficients in magnetized hot and dense electron-ion plasma relevant in binary neutron star merger simulation. We calculate electrical and thermal conductivities in low density, high temperature, highly magnetized plasma of binary neutron star mergers where quantum oscillatory behavior of electrons emerge. For pronounced thermodynamic effects, we consider zeroth Landau level population of electrons for the calculation of conductivity. We solve Boltzmann equation in presence of magnetic field to obtain the dissipative components of electrical and thermal conductivities. The dissipative coefficients are formulated considering frequency dependent dynamical screening in the quantized electron-ion scattering rate. Numerical estimations show that the effect of dynamical screening of photon propagator on electrical and thermal conductivities is pronounced. We observe that dynamical screening reduces the maxima of both the electrical and thermal conductivities by factors of thirty one and twenty respectively leading to a reduction in the corresponding time scales of these coefficients. The common scaling factor between electrical and thermal conductivity is also observed to follow cubic relationship with temperature violating Wiedemann-Franz law., Comment: Accepted in The European Physical Journal C. arXiv admin note: substantial text overlap with arXiv:2108.11878

Generalization of neural networks is crucial for deploying them safely in the real world. Common training strategies to improve generalization involve the use of data augmentations, ensembling and model averaging. In this work, we first establish a surprisingly simple but strong benchmark for generalization which utilizes diverse augmentations within a training minibatch, and show that this can learn a more balanced distribution of features. Further, we propose Diversify-Aggregate-Repeat Training (DART) strategy that first trains diverse models using different augmentations (or domains) to explore the loss basin, and further Aggregates their weights to combine their expertise and obtain improved generalization. We find that Repeating the step of Aggregation throughout training improves the overall optimization trajectory and also ensures that the individual models have a sufficiently low loss barrier to obtain improved generalization on combining them. We shed light on our approach by casting it in the framework proposed by Shen et al. and theoretically show that it indeed generalizes better. In addition to improvements in In- Domain generalization, we demonstrate SOTA performance on the Domain Generalization benchmarks in the popular DomainBed framework as well. Our method is generic and can easily be integrated with several base training algorithms to achieve performance gains., Comment: CVPR 2023. First two authors contributed equally