Search

This paper describes spatially aware Artificial Intelligence, GeoAI, tailored for small organizations such as NGOs in resource constrained contexts where access to large datasets, expensive compute infrastructure and AI expertise may be restricted. We furthermore consider future scenarios in which resource-intensive, large geospatial models may homogenize the representation of complex landscapes, and suggest strategies to prepare for this condition., Comment: 8 pages, 6 figures

Beyond fifth-generation (B5G) networks aim to support high data rates, low-latency applications, and massive machine communications. Artificial Intelligence/Machine Learning (AI/ML) can help to improve B5G network performance and efficiency. However, dynamic service demands of B5G use cases cause AI/ML model performance degradation, resulting in Service Level Agreements (SLA) violations, over- or under-provisioning of resources, etc. Retraining is essential to address the performance degradation of the AI/ML models. Existing threshold and periodic retraining approaches have potential disadvantages, such as SLA violations and inefficient resource utilization for setting a threshold parameter in a dynamic environment. This paper proposes a novel approach that predicts when to retrain AI/ML models using Generative Artificial Intelligence. The proposed predictive approach is evaluated for a Quality of Service Prediction use case on the Open Radio Access Network (O-RAN) Software Community platform and compared to the predictive approach based on the classifier and a threshold approach. Also, a realtime dataset from the Colosseum testbed is considered to evaluate Network Slicing (NS) use case with the proposed predictive approach. The results show that the proposed predictive approach outperforms both the classifier-based predictive and threshold approaches.

Home to the nation's largest Latinx (15 million), Asian American NHPI (almost 7 million) and the 5th largest Black (2.8 million) populations, California is the nation's most diverse and populous state. This diversity is one of the state's greatest assets, yet, representation of racially minoritized leaders in California's colleges and universities remains incredibly low in comparison to the racial composition of the student population. This executive summary provides key points from "Still Left Out: How Exclusion in California's Colleges & Universities Continues to Hurt Our Values, Students, and Democracy," which examines the issue of representation in leadership across California's public colleges and universities. [This report was written with significant contributions from Michele Siqueiros.]

Home to the nation's largest Latinx (15 million), Asian American NHPI (almost 7 million) and the 5th largest Black (2.8 million) populations, California is our nation's most diverse and populous state. This diversity is one of the state's greatest assets, yet, representation of racially minoritized leaders in California's colleges and universities remains incredibly low in comparison to the racial composition of the student population. "Still Left Out" provides a follow-up analysis to our watershed 2018 original publication, "Left Out: How exclusion in California's colleges and universities hurts our values, our students, and our economy," documenting the lack of full inclusion and diversity reflective of California's population among students attending the University of California (UC), California State University (CSU), and California Community Colleges. These reports include a systemwide analysis of leadership at the UC, CSU, and community colleges, plus a detailed campus-by-campus analysis of senior leaders, faculty, and academic senate bodies. We have also included a breakdown of the racial/ethnic and gender diversity of higher education governing bodies--including the UC Board of Regents, CSU Board of Trustees, California Community Colleges Board of Governors, and the California Student Aid Commission (CSAC). For California to maintain its political leadership and economic standing as the fifth-largest global economy, it must invest in the diverse talent that can adequately educate and create equitable, student-centered environments conducive to student success. The reality is that today, California's Black, Latinx, underrepresented Asian American and NHPI, and AIAN residents and scholars are Still Left Out, limiting our state's and students' potential. [This report was written with significant contributions from Michele Siqueiros.]

Generation of VLSI layout patterns is essential for a wide range of Design For Manufacturability (DFM) studies. In this study, we investigate the potential of generative machine learning models for creating design rule legal metal layout patterns. Our results demonstrate that the proposed model can generate legal patterns in complex design rule settings and achieves a high diversity score. The designed system, with its flexible settings, supports both pattern generation with localized changes, and design rule violation correction. Our methodology is validated on Intel 18A Process Design Kit (PDK) and can produce a wide range of DRC-compliant pattern libraries with only 20 starter patterns.

Chandra Surface Thermophysical Experiment (ChaSTE) is one of the payloads flown onboard the Chandrayaan-3 lander. The objective of the experiment is in-situ investigation of thermal behaviour of outermost 100 mm layer of the lunar surface by deploying a thermal probe. The probe consists of 10 temperature sensors (Platinum RTDs) mounted at different locations along the length of the probe to measure lunar soil temperatures as a function of depth. A heater is also mounted on the probe for thermal conductivity measurements. The onboard electronics of ChaSTE has two parts, Front-End Electronics (FEE) and processing electronics (PE). The front-end electronics (FEE) card is responsible for carrying out necessary sensor signal conditioning,which includes exciting the RTD sensors,acquiring analog voltages and then converting the acquired analog signals to digital signals using an Analog to Digital Converter(ADC). The front-end card is further interfaced with the processing electronics card for digital processing and spacecraft interface.The calibration, characterisation and functional test activities of Front-End Electronics of ChaSTE were carried out with the objective of testing and ensuring proper functionality and performance.A two phase calibration process involving electronic offset correction and temperature calibration were carried out. All these activities were successfully completed and the results from them provided us with a really good understanding of the behaviour of the FEE under different thermal and electrical conditions as well as when subjected to the simulated conditions of the actual ChaSTE experiment. The performance of the ChaSTE front-end electronics was very much within the design margins and its behaviour in simulated lunar environment was as desired. The data from these activities is useful in the interpretation of the actual science data of ChaSTE., Comment: 15 pages, 14 figures

Electron correlation and long-range magnetic ordering have a significant impact on the electronic structure and physical properties of solids. Here, we investigate the electronic structure of ilmenite MnTiO$_{3}$ using room temperature photoemission spectroscopy and theoretical approaches within density functional theory (DFT), DFT+$U$ and DFT+dynamical mean field theory (DMFT). Mn 2$p$ (Ti 2$p$) core level photoemission spectra, confirming Mn$^{2+}$ (Ti$^{4+}$) oxidation state, exhibit multiple satellites which are very similar to that of MnO (TiO$_{2}$), suggesting similar strength of various interactions in this system. Valence band spectra collected at different photon energies suggest dominant Mn 3$d$ character in the highest occupied band with a wide insulating gap. DFT(+$U$) correctly predicts the experimentally observed anti-ferromagnetic (AFM) insulating ground state for MnTiO$_3$ where the requirement of a large $U$ to reproduce the experimental values of magnetic moment and band gap signifies the importance of electron correlation. Magnetically disordered paramagnetic (PM) phase could be well captured within DFT+DMFT, which provides an excellent agreement for the experimental band gap, paramagnetic moment, valence band spectra as well as dominant Mn 3$d$ character in the highest occupied band. The calculated spectral function remains largely unaffected and exhibits sharper features in the magnetically ordered AFM phase. We show that the electronic structure of MnTiO$_{3}$ in both the PM and AFM phases can be accurately described within DFT+DMFT.

In this study, we tackle the challenge of limited labeled data for low-resource languages in ASR, focusing on Hindi. Specifically, we explore pseudo-labeling, by proposing a generic framework combining multiple ideas from existing works. Our framework integrates multiple base models for transcription and evaluators for assessing audio-transcript pairs, resulting in robust pseudo-labeling for low resource languages. We validate our approach with a new benchmark, IndicYT, comprising diverse YouTube audio files from multiple content categories. Our findings show that augmenting pseudo labeled data from YouTube with existing training data leads to significant performance improvements on IndicYT, without affecting performance on out-of-domain benchmarks, demonstrating the efficacy of pseudo-labeled data in enhancing ASR capabilities for low-resource languages. The benchmark, code and models developed as a part of this work will be made publicly available.

Accurate classification of celestial objects is essential for advancing our understanding of the universe. MargNet is a recently developed deep learning-based classifier applied to SDSS DR16 dataset to segregate stars, quasars, and compact galaxies using photometric data. MargNet utilizes a stacked architecture, combining a Convolutional Neural Network (CNN) for image modelling and an Artificial Neural Network (ANN) for modelling photometric parameters. In this study, we propose enhancing MargNet's performance by incorporating attention mechanisms and Vision Transformer (ViT)-based models for processing image data. The attention mechanism allows the model to focus on relevant features and capture intricate patterns within images, effectively distinguishing between different classes of celestial objects. Additionally, we leverage ViTs, a transformer-based deep learning architecture renowned for exceptional performance in image classification tasks. We enhance the model's understanding of complex astronomical images by utilizing ViT's ability to capture global dependencies and contextual information. Our approach uses a curated dataset comprising 240,000 compact and 150,000 faint objects. The models learn classification directly from the data, minimizing human intervention. Furthermore, we explore ViT as a hybrid architecture that uses photometric features and images together as input to predict astronomical objects. Our results demonstrate that the proposed attention mechanism augmented CNN in MargNet marginally outperforms the traditional MargNet and the proposed ViT-based MargNet models. Additionally, the ViT-based hybrid model emerges as the most lightweight and easy-to-train model with classification accuracy similar to that of the best-performing attention-enhanced MargNet., Comment: 33 pages, 11 figures. Accepted for publication in APSS

Modern processors implement a decoupled front-end in the form of Fetch Directed Instruction Prefetching (FDIP) to avoid front-end stalls. FDIP is driven by the Branch Prediction Unit (BPU), relying on the BPU's accuracy and branch target tracking structures to speculatively fetch instructions into the Instruction Cache (L1I). As data center applications become more complex, their code footprints also grow, resulting in an increase in Branch Target Buffer (BTB) misses. FDIP can alleviate L1I cache misses, but when it encounters a BTB miss, the BPU may not identify the current instruction as a branch to FDIP. This can prevent FDIP from prefetching or cause it to speculate down the wrong path, further polluting the L1I cache. We observe that the vast majority, 75%, of BTB-missing, unidentified branches are actually present in instruction cache lines that FDIP has previously fetched but, these missing branches have not yet been decoded and inserted into the BTB. This is because the instruction line is decoded from an entry point (which is the target of the previous taken branch) till an exit point (the taken branch). Branch instructions present in the ignored portion of the cache line we call them "Shadow Branches". Here we present Skeia, a novel shadow branch decoding technique that identifies and decodes unused bytes in cache lines fetched by FDIP, inserting them into a Shadow Branch Buffer (SBB). The SBB is accessed in parallel with the BTB, allowing FDIP to speculate despite a BTB miss. With a minimal storage state of 12.25KB, Skeia delivers a geomean speedup of ~5.7% over an 8K-entry BTB (78KB) and ~2% versus adding an equal amount of state to the BTB across 16 front-end bound applications. Since many branches stored in the SBB are unique compared to those in a similarly sized BTB, we consistently observe greater performance gains with Skeia across all examined sizes until saturation., Comment: 13 pages, 16 figures, Submitted to ASPLOS 2025

Deep convolutional neural networks (CNNs) have achieved impressive performance in many computer vision tasks. However, their large model sizes require heavy computational resources, making pruning redundant filters from existing pre-trained CNNs an essential task in developing efficient models for resource-constrained devices. Whole-network filter pruning algorithms prune varying fractions of filters from each layer, hence providing greater flexibility. Current whole-network pruning methods are either computationally expensive due to the need to calculate the loss for each pruned filter using a training dataset, or use various heuristic / learned criteria for determining the pruning fractions for each layer. This paper proposes a two-level hierarchical approach for whole-network filter pruning which is efficient and uses the classification loss as the final criterion. The lower-level algorithm (called filter-pruning) uses a sparse-approximation formulation based on linear approximation of filter weights. We explore two algorithms: orthogonal matching pursuit-based greedy selection and a greedy backward pruning approach. The backward pruning algorithm uses a novel closed-form error criterion for efficiently selecting the optimal filter at each stage, thus making the whole algorithm much faster. The higher-level algorithm (called layer-selection) greedily selects the best-pruned layer (pruning using the filter-selection algorithm) using a global pruning criterion. We propose algorithms for two different global-pruning criteria: (1) layer-wise relative error (HBGS), and (2) final classification error (HBGTS). Our suite of algorithms outperforms state-of-the-art pruning methods on ResNet18, ResNet32, ResNet56, VGG16, and ResNext101. Our method reduces the RAM requirement for ResNext101 from 7.6 GB to 1.5 GB and achieves a 94% reduction in FLOPS without losing accuracy on CIFAR-10., Comment: Accepted in TMLR 2024

As automated vehicles (AVs) become increasingly popular, the question arises as to how cyclists will interact with such vehicles. This study investigated (1) whether cyclists spontaneously notice if a vehicle is driverless, (2) how well they perform a driver-detection task when explicitly instructed, and (3) how they carry out such tasks. Using a Wizard-of-Oz method, 37 participants cycled a designated route and encountered an AV multiple times in two experimental sessions. In Session 1, participants cycled the route uninstructed, while in Session 2, they were instructed to verbally report whether they detected the presence or absence of a driver. Additionally, we recorded the participants' gaze behaviour with eye-tracking and their responses in post-session interviews. The interviews revealed that 30% of the cyclists spontaneously mentioned the absence of a driver (Session 1), and when instructed (Session 2), they detected the absence and presence of the driver with 93% accuracy. The eye-tracking data showed that cyclists looked more frequently and longer at the vehicle in Session 2 compared to Session 1. Furthermore, participants exhibited intermittent sampling of the vehicle, and they looked in front of the vehicle when it was far away and towards the windshield region when it was closer. The post-session interviews also indicated that participants were curious, felt safe, and reported a need to receive information about the AV's driving state. In conclusion, cyclists can detect the absence of a driver in the AV, and this detection may influence their perceptions of safety. Further research is needed to explore these findings in real-world traffic conditions.

With the second quantum revolution underway, quantum-enhanced sensors are moving from laboratory demonstrations to field deployments, providing enhanced and even new capabilities. Signal processing and operational software is becoming integral parts of these emerging sensing systems to reap the benefits of this progress. This paper looks into widefield Nitrogen Vacancy Center-based magnetometry and focuses on estimating the magnetic field from the Optically Detected Magnetic Resonances (ODMR) signal, a crucial output for various applications. Mapping the shifts of ODMR signals to phase estimation, a computationally efficient approaches are proposed. Involving Fourier Transform and Filtering as pre-processing steps, the suggested approaches involve linear curve fit or complex frequency estimation based on well-known super-resolution technique Estimation of Signal Parameters via Rotational Invariant Techniques (ESPRIT). The existing methods in the quantum sensing literature take different routes based on Lorentzian fitting for determining magnetic field maps. To showcase the functionality and effectiveness of the suggested techniques, relevant results, based on experimental data are provided, which shows a significant reduction in computational time with the proposed method over existing methods, Comment: 4 pages, 3 figures, typos corrected

At first light, the Thirty Meter Telescope (TMT) near-infrared (NIR) instruments will be fed by a multiconjugate adaptive optics instrument known as the Narrow Field Infrared Adaptive Optics System (NFIRAOS). NFIRAOS will use six laser guide stars to sense atmospheric turbulence in a volume corresponding to a field of view of 2', but natural guide stars (NGSs) will be required to sense tip/tilt and focus. To achieve high sky coverage (50% at the north Galactic pole), the NFIRAOS client instruments use NIR on-instrument wavefront sensors that take advantage of the sharpening of the stars by NFIRAOS. A catalog of guide stars with NIR magnitudes as faint as 22 mag in the J band (Vega system), covering the TMT-observable sky, will be a critical resource for the efficient operation of NFIRAOS, and no such catalog currently exists. Hence, it is essential to develop such a catalog by computing the expected NIR magnitudes of stellar sources identified in deep optical sky surveys using their optical magnitudes. This paper discusses the generation of a partial NIR Guide Star Catalog (IRGSC), similar to the final IRGSC for TMT operations. The partial catalog is generated by applying stellar atmospheric models to the optical data of stellar sources from the Panoramic Survey Telescope and Rapid Response System (Pan-STARRS) optical data and then computing their expected NIR magnitudes. We validated the computed NIR magnitudes of the sources in some fields by using the available NIR data for those fields. We identified the remaining challenges of this approach. We outlined the path for producing the final IRGSC using the Pan-STARRS data. We have named the Python code to generate the IRGSC as irgsctool, which generates a list of NGS for a field using optical data from the Pan-STARRS 3pi survey and also a list of NGSs having observed NIR data from the UKIRT Infrared Deep Sky Survey if they are available.

This paper presents results from Kepler photometric light curves and high-resolution spectroscopic study of a super Li-rich giant KIC11087027. Using the light curve analysis, we measured the star's rotational period P$_{\rm rot}$=30.4$\pm$0.1~days, which translates to rotational velocity V$_{\rm rot}$=19.5 $\pm$ 1.7~km s$^{-1}$. Star's location in the HR-diagram, derived values of $^{12}C/^{13}C$ = 7$\pm$1 and $[C/N]=-0.95\pm 0.2$, and the inferred asteroseismic parameters from secondary calibration based on spectra suggest star is a low-mass red clump giant in the He-core burning phase. Using Gaia data, we found evidence of variation in radial velocity and proper motion, indicative of presence of an unresolved binary. The large V$_{\rm rot}$ is probably a result of tidal synchronization combined with the after-effects of He-flash, in which the size of the star is reduced significantly. The simultaneous presence of features like high rotation, very high Li abundance, strong dust shell, and strong flares in a single star is relatively uncommon, suggesting that the star experiencing tidal synchronization has recently undergone He-flash. The results pose a question whether the binary interaction, hence the high rotation, is a prerequisite for dredging-up of the high amounts of Li from the interior to the photosphere during or immediately after the He-flash event., Comment: 9 pages, 3 figures, 2 tables, accepted

Modern OOO CPUs have very deep pipelines with large branch misprediction recovery penalties. Speculatively executed instructions on the wrong path can significantly change cache state, depending on speculation levels. Architects often employ trace-driven simulation models in the design exploration stage, which sacrifice precision for speed. Trace-driven simulators are orders of magnitude faster than execution-driven models, reducing the often hundreds of thousands of simulation hours needed to explore new micro-architectural ideas. Despite this strong benefit of trace-driven simulation, these often fail to adequately model the consequences of wrong path because obtaining them is nontrivial. Prior works consider either a positive or negative impact of wrong path but not both. Here, we examine wrong path execution in simulation results and design a set of infrastructure for enabling wrong-path execution in a trace driven simulator. Our analysis shows the wrong path affects structures on both the instruction and data sides extensively, resulting in performance variations ranging from $-3.05$\% to $20.9$\% when ignoring wrong path. To benefit the research community and enhance the accuracy of simulators, we opened our traces and tracing utility in the hopes that industry can provide wrong-path traces generated by their internal simulators, enabling academic simulation without exposing industry IP., Comment: 5 pages, 7 Figures, Submited to Computer Architecture Letters

In this paper, we propose a novel encoder-decoder architecture, named SABER, to learn the 6D pose of the object in the embedding space by learning shape representation at a given pose. This model enables us to learn pose by performing shape representation at a target pose from RGB image input. We perform shape representation as an auxiliary task which helps us in learning rotations space for an object based on 2D images. An image encoder predicts the rotation in the embedding space and the DeepSDF based decoder learns to represent the object's shape at the given pose. As our approach is shape based, the pipeline is suitable for any type of object irrespective of the symmetry. Moreover, we need only a CAD model of the objects to train SABER. Our pipeline is synthetic data based and can also handle symmetric objects without symmetry labels and, thus, no additional labeled training data is needed. The experimental evaluation shows that our method achieves close to benchmark results for both symmetric objects and asymmetric objects on Occlusion-LineMOD, and T-LESS datasets., Comment: ECCV 2024 R6D workshop

In memory computing (IMC) architectures for deep learning (DL) accelerators leverage energy-efficient and highly parallel matrix vector multiplication (MVM) operations, implemented directly in memory arrays. Such IMC designs have been explored based on CMOS as well as emerging non-volatile memory (NVM) technologies like RRAM. IMC architectures generally involve a large number of cores consisting of memory arrays, storing the trained weights of the DL model. Peripheral units like DACs and ADCs are also used for applying inputs and reading out the output values. Recently reported designs reveal that the ADCs required for reading out the MVM results, consume more than 85% of the total compute power and also dominate the area, thereby eschewing the benefits of the IMC scheme. Mitigation of imperfections in the ADCs, namely, non-linearity and variations, incur significant design overheads, due to dedicated calibration units. In this work we present peripheral aware design of IMC cores, to mitigate such overheads. It involves incorporating the non-idealities of ADCs in the training of the DL models, along with that of the memory units. The proposed approach applies equally well to both current mode as well as charge mode MVM operations demonstrated in recent years., and can significantly simplify the design of mixed-signal IMC units.

We use JWST/NIRSpec observations from the Assembly of Ultradeep Rest-optical Observations Revealing Astrophysics (AURORA) survey to constrain the shape of the nebular attenuation curve of a star-forming galaxy at z=4.41, GOODSN-17940. We utilize 11 unblended HI recombination lines to derive the attenuation curve spanning optical to near-infrared wavelengths (3751-9550 \r{A}). We then leverage a high-S/N spectroscopic detection of the rest-frame ultraviolet continuum in combination with rest-UV photometric measurements to constrain the shape of the curve at ultraviolet wavelengths. While this UV constraint is predominantly based on stellar emission, the large measured equivalent widths of H$\alpha$ and H$\beta$ indicate that GOODSN-17940 is dominated by an extremely young stellar population <10 Myr in age such that the UV stellar continuum experiences the same attenuation as the nebular emission. The resulting combined nebular attenuation curve spans 1400-9550 \r{A} and has a shape that deviates significantly from commonly assumed dust curves in high-redshift studies. Relative to the Milky Way, SMC, and Calzetti curves, the new curve has a steeper slope at long wavelengths ($\lambda>5000$ \r{A}) while displaying a similar slope across blue-optical wavelengths ($\lambda=3750-5000$ \r{A}). In the ultraviolet, the new curve is shallower than the SMC and Calzetti curves and displays no significant 2175 \r{A} bump. This work demonstrates that the most commonly assumed dust curves are not appropriate for all high-redshift galaxies. These results highlight the ability to derive nebular attenuation curves for individual high-redshift sources with deep JWST/NIRSpec spectroscopy, thereby improving the accuracy of physical properties inferred from nebular emission lines., Comment: 21 pages, 6 figures, 1 table, submitted to ApJ

Multilingual speaker verification introduces the challenge of verifying a speaker in multiple languages. Existing systems were built using i-vector/x-vector approaches along with Bi-LSTMs, which were trained to discriminate speakers, irrespective of the language. Instead of exploring the design space manually, we propose a neural architecture search for multilingual speaker verification suitable for mobile devices, called \textbf{NeuralMultiling}. First, our algorithm searches for an optimal operational combination of neural cells with different architectures for normal cells and reduction cells and then derives a CNN model by stacking neural cells. Using the derived architecture, we performed two different studies:1) language agnostic condition and 2) interoperability between languages and devices on the publicly available Multilingual Audio-Visual Smartphone (MAVS) dataset. The experimental results suggest that the derived architecture significantly outperforms the existing Autospeech method by a 5-6\% reduction in the Equal Error Rate (EER) with fewer model parameters.

We introduce the concept of Hybrid Surrogate Models (HSMs) -- combining multivariate polynomials with Heavyside functions -- as approximates of functions with finitely many jump discontinuities. We exploit the HSMs for formulating a variational optimization approach, solving non-regular partial differential equations (PDEs) with non-continuous shock-type solutions. The HSM technique simultaneously obtains a parametrization of the position and the height of the shocks as well as the solution of the PDE. We show that the HSM technique circumvents the notorious Gibbs phenomenon, which limits the accuracy that classic numerical methods reach. Numerical experiments, addressing linear and non-linearly propagating shocks, demonstrate the strong approximation power of the HSM technique.

The ability to quantify how well an action is carried out, also known as action quality assessment (AQA), has attracted recent interest in the vision community. Unfortunately, prior methods often ignore the score rubric used by human experts and fall short of quantifying the uncertainty of the model prediction. To bridge the gap, we present RICA^2 - a deep probabilistic model that integrates score rubric and accounts for prediction uncertainty for AQA. Central to our method lies in stochastic embeddings of action steps, defined on a graph structure that encodes the score rubric. The embeddings spread probabilistic density in the latent space and allow our method to represent model uncertainty. The graph encodes the scoring criteria, based on which the quality scores can be decoded. We demonstrate that our method establishes new state of the art on public benchmarks, including FineDiving, MTL-AQA, and JIGSAWS, with superior performance in score prediction and uncertainty calibration. Our code is available at https://abrarmajeedi.github.io/rica2_aqa/, Comment: Accepted at European Conference on Computer Vision (ECCV) 2024

LLM-driven dialog systems are used in a diverse set of applications, ranging from healthcare to customer service. However, given their generalization capability, it is difficult to ensure that these chatbots stay within the boundaries of the specialized domains, potentially resulting in inaccurate information and irrelevant responses. This paper introduces an unsupervised approach for automatically inducing domain-specific dialog flows that can be used to constrain LLM-based chatbots. We introduce two variants of dialog flow based on the availability of in-domain conversation instances. Through human and automatic evaluation over various dialog domains, we demonstrate that our high-quality data-guided dialog flows achieve better domain coverage, thereby overcoming the need for extensive manual crafting of such flows., Comment: Accepted at SIGDIAL 2024

Depth estimation is a critical topic for robotics and vision-related tasks. In monocular depth estimation, in comparison with supervised learning that requires expensive ground truth labeling, self-supervised methods possess great potential due to no labeling cost. However, self-supervised learning still has a large gap with supervised learning in 3D reconstruction and depth estimation performance. Meanwhile, scaling is also a major issue for monocular unsupervised depth estimation, which commonly still needs ground truth scale from GPS, LiDAR, or existing maps to correct. In the era of deep learning, existing methods primarily rely on exploring image relationships to train unsupervised neural networks, while the physical properties of the camera itself such as intrinsics and extrinsics are often overlooked. These physical properties are not just mathematical parameters; they are embodiments of the camera's interaction with the physical world. By embedding these physical properties into the deep learning model, we can calculate depth priors for ground regions and regions connected to the ground based on physical principles, providing free supervision signals without the need for additional sensors. This approach is not only easy to implement but also enhances the effects of all unsupervised methods by embedding the camera's physical properties into the model, thereby achieving an embodied understanding of the real world.

To address the rural broadband challenge and to leverage the unique opportunities that rural regions provide for piloting advanced wireless applications, we design and implement the ARA wireless living lab for research and innovation in rural wireless systems and their applications in precision agriculture, community services, and so on. ARA focuses on the unique community, application, and economic context of rural regions, and it features the first-of-its-kind, real-world deployment of long-distance, high-capacity wireless x-haul and access platforms across a rural area of diameter over 30 km. With both software-defined radios and programmable COTS systems and through effective orchestration of these wireless resources with fiber as well as compute resources embedded end-to-end across user equipment, base stations, edge, and cloud, ARA offers programmability, performance, robustness, and heterogeneity at the same time, thus enabling rural-focused co-evolution of wireless and applications while helping advance the frontiers of wireless systems in domains such as O-RAN, NextG, and agriculture applications. Here we present the design principles and implementation strategies of ARA, characterize its performance and heterogeneity, and highlight example wireless and application experiments uniquely enabled by ARA., Comment: 17 pages, 18 figures

Vehicular Ad-Hoc Networks (VANETs) hold immense potential for improving traffic safety and efficiency. However, traditional centralized approaches for machine learning in VANETs raise concerns about data privacy and security. Federated Learning (FL) offers a solution that enables collaborative model training without sharing raw data. This paper proposes FL-DECO-BC as a novel privacy-preserving, provably secure, and provenance-preserving federated learning framework specifically designed for VANETs. FL-DECO-BC leverages decentralized oracles on blockchain to securely access external data sources while ensuring data privacy through advanced techniques. The framework guarantees provable security through cryptographic primitives and formal verification methods. Furthermore, FL-DECO-BC incorporates a provenance-preserving design to track data origin and history, fostering trust and accountability. This combination of features empowers VANETs with secure and privacy-conscious machine-learning capabilities, paving the way for advanced traffic management and safety applications.

Analyses of the cosmic 21-cm signal are hampered by astrophysical foregrounds that are far stronger than the signal itself. These foregrounds, typically confined to a wedge-shaped region in Fourier space, often necessitate the removal of a vast majority of modes, thereby degrading the quality of the data anisotropically. To address this challenge, we introduce a novel deep generative model based on stochastic interpolants to reconstruct the 21-cm data lost to wedge filtering. Our method leverages the non-Gaussian nature of the 21-cm signal to effectively map wedge-filtered 3D lightcones to samples from the conditional distribution of wedge-recovered lightcones. We demonstrate how our method is able to restore spatial information effectively, considering both varying cosmological initial conditions and astrophysics. Furthermore, we discuss a number of future avenues where this approach could be applied in analyses of the 21-cm signal, potentially offering new opportunities to improve our understanding of the Universe during the epochs of cosmic dawn and reionization., Comment: 12 pages, 5 figures

Recent work [B. Fore and S. Reddy, Phys. Rev. C 101 035809 (2020)] has shown that the population of thermal pions could modify the equation of state and transport properties of hot and dense neutron-rich matter and introduce new reaction pathways to change the proton fraction. In this article we study their impact on the bulk viscosity of dense matter, focusing on the neutrino-trapped regime that would be realized in neutron star mergers and supernovae. We find that the presence of a thermal population of pions alters the bulk viscosity by modifying the EoS (via the susceptibilities) and by providing new reaction pathways to achieve beta-equilibrium. In neutron star merger conditions, the bulk viscosity in neutrino-trapped $npe\mu$ matter (without pions) has its peak at temperatures of at most a couple MeV and is quite small at temperatures of tens of MeV. We find that thermal pions enhance the low-temperature peak of the bulk viscosity by a factor of a few and shift it to slightly lower temperatures. At higher temperatures, where the pion abundance is large but the bulk viscosity is traditionally small, pions can increase the bulk viscosity by an order of magnitude or more, although it is still orders of magnitude smaller than its peak value.

Modeling human cognitive processes in dynamic decision-making tasks has been an endeavor in AI for a long time because such models can help make AI systems more intuitive, personalized, mitigate any human biases, and enhance training in simulation. Some initial work has attempted to utilize neural networks (and large language models) but often assumes one common model for all humans and aims to emulate human behavior in aggregate. However, the behavior of each human is distinct, heterogeneous, and relies on specific past experiences in certain tasks. For instance, consider two individuals responding to a phishing email: one who has previously encountered and identified similar threats may recognize it quickly, while another without such experience might fall for the scam. In this work, we build on Instance Based Learning (IBL) that posits that human decisions are based on similar situations encountered in the past. However, IBL relies on simple fixed form functions to capture the mapping from past situations to current decisions. To that end, we propose two new attention-based neural network models to have open form non-linear functions to model distinct and heterogeneous human decision-making in dynamic settings. We experiment with two distinct datasets gathered from human subject experiment data, one focusing on detection of phishing email by humans and another where humans act as attackers in a cybersecurity setting and decide on an attack option. We conducted extensive experiments with our two neural network models, IBL, and GPT3.5, and demonstrate that the neural network models outperform IBL significantly in representing human decision-making, while providing similar interpretability of human decisions as IBL. Overall, our work yields promising results for further use of neural networks in cognitive modeling of human decision making., Comment: Our code is available at https://github.com/shshnkreddy/NCM-HDM

Video Question Answering is a challenging task, which requires the model to reason over multiple frames and understand the interaction between different objects to answer questions based on the context provided within the video, especially in datasets like NExT-QA (Xiao et al., 2021a) which emphasize on causal and temporal questions. Previous approaches leverage either sub-sampled information or causal intervention techniques along with complete video features to tackle the NExT-QA task. In this work we elicit the limitations of these approaches and propose solutions along four novel directions of improvements on theNExT-QA dataset. Our approaches attempts to compensate for the shortcomings in the previous works by systematically attacking each of these problems by smartly sampling frames, explicitly encoding actions and creating interventions that challenge the understanding of the model. Overall, for both single-frame (+6.3%) and complete-video (+1.1%) based approaches, we obtain the state-of-the-art results on NExT-QA dataset.

Virtual scenario-based testing methods to validate autonomous driving systems are predominantly centred around collision avoidance, and lack a comprehensive approach to evaluate optimal driving behaviour holistically. Furthermore, current validation approaches do not align with authorisation and monitoring requirements put forth by regulatory bodies. We address these validation gaps by outlining a universal evaluation framework that: incorporates the notion of careful and competent driving, unifies behavioural competencies and evaluation criteria, and is amenable at a scenario-specific and aggregate behaviour level. This framework can be leveraged to evaluate optimal driving in scenario-based testing, and for post-deployment monitoring to ensure continual compliance with regulation and safety standards., 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

Given a graph $G = (V, E)$ and an integer $k$, the Minimum Membership Dominating Set problem asks to compute a set $S \subseteq V$ such that for each $v \in V$, $1 \leq |N[v] \cap S| \leq k$. The problem is known to be NP-complete even on split graphs and planar bipartite graphs. In this paper, we approach the problem from the algorithmic standpoint and obtain several interesting results. We give an $\mathcal{O}^*(1.747^n)$ time algorithm for the problem on split graphs. Following a reduction from a special case of 1-in-3 SAT problem, we show that there is no sub-exponential time algorithm running in time $\mathcal{O}^*(2^{o(n)})$ for bipartite graphs, for any $k \geq 2$. We also prove that the problem is NP-complete when $\Delta = k+2$, for any $k\geq 5$, even for bipartite graphs. We investigate the parameterized complexity of the problem for the parameter twin cover and the combined parameter distance to cluster, membership($k$) and prove that the problem is fixed-parameter tractable. Using a dynamic programming based approach, we obtain a linear-time algorithm for trees., Comment: 23 pages, 6 figures

During social interactions, understanding the intricacies of the context can be vital, particularly for socially anxious individuals. While previous research has found that the presence of a social interaction can be detected from ambient audio, the nuances within social contexts, which influence how anxiety provoking interactions are, remain largely unexplored. As an alternative to traditional, burdensome methods like self-report, this study presents a novel approach that harnesses ambient audio segments to detect social threat contexts. We focus on two key dimensions: number of interaction partners (dyadic vs. group) and degree of evaluative threat (explicitly evaluative vs. not explicitly evaluative). Building on data from a Zoom-based social interaction study (N=52 college students, of whom the majority N=45 are socially anxious), we employ deep learning methods to achieve strong detection performance. Under sample-wide 5-fold Cross Validation (CV), our model distinguished dyadic from group interactions with 90\% accuracy and detected evaluative threat at 83\%. Using a leave-one-group-out CV, accuracies were 82\% and 77\%, respectively. While our data are based on virtual interactions due to pandemic constraints, our method has the potential to extend to diverse real-world settings. This research underscores the potential of passive sensing and AI to differentiate intricate social contexts, and may ultimately advance the ability of context-aware digital interventions to offer personalized mental health support., Comment: 8 pages, 4 figures, 3 tables. Accepted by ACII 2024, Glasgow, UK. To appear in the Proceedings of ACII 2024

Scams are a widespread issue with severe consequences for both victims and perpetrators, but existing data collection is fragmented, precluding global and comparative local understanding. The present study addresses this gap through a nationally representative survey (n = 8,369) on scam exposure, victimization, types, vectors, and reporting in 12 countries: Belgium, Egypt, France, Hungary, Indonesia, Mexico, Romania, Slovakia, South Africa, South Korea, Sweden, and the United Kingdom. We analyze 6 survey questions to build a detailed quantitative picture of the scams landscape in each country, and compare across countries to identify global patterns. We find, first, that residents of less affluent countries suffer financial loss from scams more often. Second, we find that the internet plays a key role in scams across the globe, and that GNI per-capita is strongly associated with specific scam types and contact vectors. Third, we find widespread under-reporting, with residents of less affluent countries being less likely to know how to report a scam. Our findings contribute valuable insights for researchers, practitioners, and policymakers in the online fraud and scam prevention space., Comment: To appear in the Journal of Online Trust and Safety

Almost in every heavily computation-dependent application, from 6G communication systems to autonomous driving platforms, a large portion of computing should be near to the client side. Edge computing (AI at Edge) in mobile devices is one of the optimized approaches for addressing this requirement. Therefore, in this work, the possibilities and challenges of implementing a low-latency and power-optimized smart mobile system are examined. Utilizing Field Programmable Gate Array (FPGA) based solutions at the edge will lead to bandwidth-optimized designs and as a consequence can boost the computational effectiveness at a system-level deadline. Moreover, various performance aspects and implementation feasibilities of Neural Networks (NNs) on both embedded FPGA edge devices (using Xilinx Multiprocessor System on Chip (MPSoC)) and Cloud are discussed throughout this research. The main goal of this work is to demonstrate a hybrid system that uses the deep learning programmable engine developed by Xilinx Inc. as the main component of the hardware accelerator. Then based on this design, an efficient system for mobile edge computing is represented by utilizing an embedded solution., Comment: 25. ITG Fachtagung Mobilkommunikation

Stuttering is a common speech impediment that is caused by irregular disruptions in speech production, affecting over 70 million people across the world. Standard automatic speech processing tools do not take speech ailments into account and are thereby not able to generate meaningful results when presented with stuttered speech as input. The automatic detection of stuttering is an integral step towards building efficient, context-aware speech processing systems. While previous approaches explore both statistical and neural approaches for stuttering detection, all of these methods are uni-modal in nature. This paper presents MMSD-Net, the first multi-modal neural framework for stuttering detection. Experiments and results demonstrate that incorporating the visual signal significantly aids stuttering detection, and our model yields an improvement of 2-17% in the F1-score over existing state-of-the-art uni-modal approaches., Comment: Accepted at INTERSPEECH 2024

Foundation models and vision-language pre-training have notably advanced Vision Language Models (VLMs), enabling multimodal processing of visual and linguistic data. However, their performance has been typically assessed on general scene understanding - recognizing objects, attributes, and actions - rather than cultural comprehension. This study introduces CulturalVQA, a visual question-answering benchmark aimed at assessing VLM's geo-diverse cultural understanding. We curate a collection of 2,378 image-question pairs with 1-5 answers per question representing cultures from 11 countries across 5 continents. The questions probe understanding of various facets of culture such as clothing, food, drinks, rituals, and traditions. Benchmarking VLMs on CulturalVQA, including GPT-4V and Gemini, reveals disparity in their level of cultural understanding across regions, with strong cultural understanding capabilities for North America while significantly lower performance for Africa. We observe disparity in their performance across cultural facets too, with clothing, rituals, and traditions seeing higher performances than food and drink. These disparities help us identify areas where VLMs lack cultural understanding and demonstrate the potential of CulturalVQA as a comprehensive evaluation set for gauging VLM progress in understanding diverse cultures.

Preserving the privacy of preferences (or rewards) of a sequential decision-making agent when decisions are observable is crucial in many physical and cybersecurity domains. For instance, in wildlife monitoring, agents must allocate patrolling resources without revealing animal locations to poachers. This paper addresses privacy preservation in planning over a sequence of actions in MDPs, where the reward function represents the preference structure to be protected. Observers can use Inverse RL (IRL) to learn these preferences, making this a challenging task. Current research on differential privacy in reward functions fails to ensure guarantee on the minimum expected reward and offers theoretical guarantees that are inadequate against IRL-based observers. To bridge this gap, we propose a novel approach rooted in the theory of deception. Deception includes two models: dissimulation (hiding the truth) and simulation (showing the wrong). Our first contribution theoretically demonstrates significant privacy leaks in existing dissimulation-based methods. Our second contribution is a novel RL-based planning algorithm that uses simulation to effectively address these privacy concerns while ensuring a guarantee on the expected reward. Experiments on multiple benchmark problems show that our approach outperforms previous methods in preserving reward function privacy., Comment: ECAI 2024

Errors or failures in a high-volume manufacturing environment can have significant impact that can result in both the loss of time and money. Identifying such failures early has been a top priority for manufacturing industries and various rule-based algorithms have been developed over the years. However, catching these failures is time consuming and such algorithms cannot adapt well to changes in designs, and sometimes variations in everyday behavior. More importantly, the number of units to monitor in a high-volume manufacturing environment is too big for manual monitoring or for a simple program. Here we develop a novel program that combines both rule-based decisions and machine learning models that can not only learn and adapt to such day-to-day variations or long-term design changes, but also can be applied at scale to the high number of manufacturing units in use today. Using the current state-of-the-art technologies, we then deploy this program at-scale to handle the needs of ever-increasing demand from the manufacturing environment.

Enterprise chatbots, powered by generative AI, are emerging as key applications to enhance employee productivity. Retrieval Augmented Generation (RAG), Large Language Models (LLMs), and orchestration frameworks like Langchain and Llamaindex are crucial for building these chatbots. However, creating effective enterprise chatbots is challenging and requires meticulous RAG pipeline engineering. This includes fine-tuning embeddings and LLMs, extracting documents from vector databases, rephrasing queries, reranking results, designing prompts, honoring document access controls, providing concise responses, including references, safeguarding personal information, and building orchestration agents. We present a framework for building RAG-based chatbots based on our experience with three NVIDIA chatbots: for IT/HR benefits, financial earnings, and general content. Our contributions are three-fold: introducing the FACTS framework (Freshness, Architectures, Cost, Testing, Security), presenting fifteen RAG pipeline control points, and providing empirical results on accuracy-latency tradeoffs between large and small LLMs. To the best of our knowledge, this is the first paper of its kind that provides a holistic view of the factors as well as solutions for building secure enterprise-grade chatbots.", Comment: 8 pages, 6 figures, 2 tables, Preprint submission to ACM CIKM 2024

Model training requires significantly more memory, compared with inference. Parameter efficient fine-tuning (PEFT) methods provide a means of adapting large models to downstream tasks using less memory. However, existing methods such as adapters, prompt tuning or low-rank adaptation (LoRA) either introduce latency overhead at inference time or achieve subpar downstream performance compared with full fine-tuning. In this work we propose Random Subspace Adaptation (ROSA), a method that outperforms previous PEFT methods by a significant margin, while maintaining a zero latency overhead during inference time. In contrast to previous methods, ROSA is able to adapt subspaces of arbitrarily large dimension, better approximating full-finetuning. We demonstrate both theoretically and experimentally that this makes ROSA strictly more expressive than LoRA, without consuming additional memory during runtime. As PEFT methods are especially useful in the natural language processing domain, where models operate on scales that make full fine-tuning very expensive, we evaluate ROSA in two common NLP scenarios: natural language generation (NLG) and natural language understanding (NLU) with GPT-2 and RoBERTa, respectively. We show that on almost every GLUE task ROSA outperforms LoRA by a significant margin, while also outperforming LoRA on NLG tasks. Our code is available at https://github.com/rosa-paper/rosa