arXiv subjects

Kevin Chen

Publications and source records attributed to Kevin Chen.

At least 37 records · Page 2Linked to original sources

arXiv2025

The global landscape of academic guidelines for generative AI and Large Language Models

The integration of Generative Artificial Intelligence (GAI) and Large Language Models (LLMs) in academia has spurred a global discourse on their potential pedagogical benefits and ethical considerations. Positive reactions highlight some potential, such as collaborative creativity, increased access to education, and empowerment of trainers and trainees. However, negative reactions raise concerns about ethical complexities, balancing innovation and academic integrity, unequal access, and misinformation risks. Through a systematic survey and text-mining-based analysis of global and national directives, insights from independent research, and eighty university-level guidelines, this study provides a nuanced understanding of the opportunities and challenges posed by GAI and LLMs in education. It emphasizes the importance of balanced approaches that harness the benefits of these technologies while addressing ethical considerations and ensuring equitable access and educational outcomes. The paper concludes with recommendations for fostering responsible innovation and ethical practices to guide the integration of GAI and LLMs in academia.

cs.CY↗

arXiv2025

IGGA: A Dataset of Industrial Guidelines and Policy Statements for Generative AIs

This paper introduces IGGA, a dataset of 160 industry guidelines and policy statements for the use of Generative AIs (GAIs) and Large Language Models (LLMs) in industry and workplace settings, collected from official company websites, and trustworthy news sources. The dataset contains 104,565 words and serves as a valuable resource for natural language processing tasks commonly applied in requirements engineering, such as model synthesis, abstraction identification, and document structure assessment. Additionally, IGGA can be further annotated to function as a benchmark for various tasks, including ambiguity detection, requirements categorization, and the identification of equivalent requirements. Our methodologically rigorous approach ensured a thorough examination, with a selection of reputable and influential companies that represent a diverse range of global institutions across six continents. The dataset captures perspectives from fourteen industry sectors, including technology, finance, and both public and private institutions, offering a broad spectrum of insights into the integration of GAIs and LLMs in industry.

cs.CY↗

arXiv2025

AGGA: A Dataset of Academic Guidelines for Generative AI and Large Language Models

This study introduces AGGA, a dataset comprising 80 academic guidelines for the use of Generative AIs (GAIs) and Large Language Models (LLMs) in academic settings, meticulously collected from official university websites. The dataset contains 188,674 words and serves as a valuable resource for natural language processing tasks commonly applied in requirements engineering, such as model synthesis, abstraction identification, and document structure assessment. Additionally, AGGA can be further annotated to function as a benchmark for various tasks, including ambiguity detection, requirements categorization, and the identification of equivalent requirements. Our methodologically rigorous approach ensured a thorough examination, with a selection of universities that represent a diverse range of global institutions, including top-ranked universities across six continents. The dataset captures perspectives from a variety of academic fields, including humanities, technology, and both public and private institutions, offering a broad spectrum of insights into the integration of GAIs and LLMs in academia.

cs.CL↗

arXiv2025

Scalable low loss cryogenic packaging of quantum memories in CMOS-foundry processed photonic chips

Optically linked solid-state quantum memories such as color centers in diamond are a promising platform for distributed quantum information processing and networking. Photonic integrated circuits (PICs) have emerged as a crucial enabling technology for these systems, integrating quantum memories with efficient electrical and optical interfaces in a compact and scalable platform. Packaging these hybrid chips into deployable modules while maintaining low optical loss and resiliency to temperature cycling is a central challenge to their practical use. We demonstrate a packaging method for PICs using surface grating couplers and angle-polished fiber arrays that is robust to temperature cycling, offers scalable channel count, applies to a wide variety of PIC platforms and wavelengths, and offers pathways to automated high-throughput packaging. Using this method, we show optically and electrically packaged quantum memory modules integrating all required qubit controls on chip, operating at millikelvin temperatures with < 3dB losses achievable from fiber to quantum memory.

quant-ph↗

arXiv2025

Reinforcement Learning for Long-Horizon Interactive LLM Agents

Interactive digital agents (IDAs) leverage APIs of stateful digital environments to perform tasks in response to user requests. While IDAs powered by instruction-tuned large language models (LLMs) can react to feedback from interface invocations in multi-step exchanges, they have not been trained in their respective digital environments. Prior methods accomplish less than half of tasks in sophisticated benchmarks such as AppWorld. We present a reinforcement learning (RL) approach that trains IDAs directly in their target environments. We formalize this training as a partially observable Markov decision process and derive LOOP, a data- and memory-efficient variant of proximal policy optimization. LOOP uses no value network and maintains exactly one copy of the underlying LLM in memory, making its implementation straightforward and as memory-efficient as fine-tuning a single LLM. A 32-billion-parameter agent trained with LOOP in the AppWorld environment outperforms the much larger OpenAI o1 agent by 9 percentage points (15% relative). To our knowledge, this is the first reported application of RL to IDAs that interact with a stateful, multi-domain, multi-app environment via direct API calls. Our analysis sheds light on the effectiveness of RL in this area, showing that the agent learns to consult the API documentation, avoid unwarranted assumptions, minimize confabulation, and recover from setbacks.

cs.LG↗

arXiv2024

An Empty Room is All We Want: Automatic Defurnishing of Indoor Panoramas

We propose a pipeline that leverages Stable Diffusion to improve inpainting results in the context of defurnishing -- the removal of furniture items from indoor panorama images. Specifically, we illustrate how increased context, domain-specific model fine-tuning, and improved image blending can produce high-fidelity inpaints that are geometrically plausible without needing to rely on room layout estimation. We demonstrate qualitative and quantitative improvements over other furniture removal techniques.

cs.CV↗

arXiv2024

Stereo-NEC: Enhancing Stereo Visual-Inertial SLAM Initialization with Normal Epipolar Constraints

We propose an accurate and robust initialization approach for stereo visual-inertial SLAM systems. Unlike the current state-of-the-art method, which heavily relies on the accuracy of a pure visual SLAM system to estimate inertial variables without updating camera poses, potentially compromising accuracy and robustness, our approach offers a different solution. We realize the crucial impact of precise gyroscope bias estimation on rotation accuracy. This, in turn, affects trajectory accuracy due to the accumulation of translation errors. To address this, we first independently estimate the gyroscope bias and use it to formulate a maximum a posteriori problem for further refinement. After this refinement, we proceed to update the rotation estimation by performing IMU integration with gyroscope bias removed from gyroscope measurements. We then leverage robust and accurate rotation estimates to enhance translation estimation via 3-DoF bundle adjustment. Moreover, we introduce a novel approach for determining the success of the initialization by evaluating the residual of the normal epipolar constraint. Extensive evaluations on the EuRoC dataset illustrate that our method excels in accuracy and robustness. It outperforms ORB-SLAM3, the current leading stereo visual-inertial initialization method, in terms of absolute trajectory error and relative rotation error, while maintaining competitive computational speed. Notably, even with 5 keyframes for initialization, our method consistently surpasses the state-of-the-art approach using 10 keyframes in rotation accuracy.

cs.RO↗

arXiv2024

Bringing Social Computing to Secondary School Classrooms

Social computing is the study of how technology shapes human social interactions. This topic has become increasingly relevant to secondary school students (ages 11--18) as more of young people's everyday social experiences take place online, particularly with the continuing effects of the COVID-19 pandemic. However, social computing topics are rarely touched upon in existing middle and high school curricula. We seek to introduce concepts from social computing to secondary school students so they can understand how computing has wide-ranging social implications that touch upon their everyday lives, as well as think critically about both the positive and negative sides of different social technology designs. In this report, we present a series of six lessons combining presentations and hands-on activities covering topics within social computing and detail our experience teaching these lessons to approximately 1,405 students across 13 middle and high schools in our local school district. We developed lessons covering how social computing relates to the topics of Data Management, Encrypted Messaging, Human-Computer Interaction Careers, Machine Learning and Bias, Misinformation, and Online Behavior. We found that 81.13% of students expressed greater interest in the content of our lessons compared to their interest in STEM overall. We also found from pre- and post-lesson comprehension questions that 63.65% learned new concepts from the main activity. We release all lesson materials on a website for public use. From our experience, we observed that students were engaged in these topics and found enjoyment in finding connections between computing and their own lives.

cs.CY↗

arXiv2023

Nanoelectromechanical control of spin-photon interfaces in a hybrid quantum system on chip

Atom-like defects or color centers (CC's) in nanostructured diamond are a leading platform for optically linked quantum technologies, with recent advances including memory-enhanced quantum communication, multi-node quantum networks, and spin-mediated generation of photonic cluster states. Scaling to practically useful applications motivates architectures meeting the following criteria: C1 individual optical addressing of spin qubits; C2 frequency tuning of CC spin-dependent optical transitions; C3 coherent spin control in CC ground states; C4 active photon routing; C5 scalable manufacturability; and C6 low on-chip power dissipation for cryogenic operations. However, no architecture meeting C1-C6 has thus far been demonstrated. Here, we introduce a hybrid quantum system-on-chip (HQ-SoC) architecture that simultaneously achieves C1-C6. Key to this advance is the realization of piezoelectric strain control of diamond waveguide-coupled tin vacancy centers to meet C2 and C3, with ultra-low power dissipation necessary for C6. The DC response of our device allows emitter transition tuning by over 20 GHz, while the large frequency range (exceeding 2 GHz) enables low-power AC control. We show acoustic manipulation of integrated tin vacancy spins and estimate single-phonon coupling rates over 1 kHz in the resolved sideband regime. Combined with high-speed optical routing with negligible static hold power, this HQ-SoC platform opens the path to scalable single-qubit control with optically mediated entangling gates.

quant-ph↗

arXiv2022

Janus and RG-flow interfaces in three-dimensional gauged supergravity

In this paper, we construct Janus-type solutions of three-dimensional gauged supergravity with sixteen supersymmetries. We find solutions which correspond to interfaces between the same CFT on both sides, as well as RG-flow interfaces between CFTs with different numbers of supersymmetries and central charges. The solutions are obtained by solving the flow equations derived from the supersymmetry variations, and they preserve some fraction of the supersymmetries of the $\AdS_3$ vacua.

hep-th↗

arXiv2021

Learning Language-Conditioned Robot Behavior from Offline Data and Crowd-Sourced Annotation

We study the problem of learning a range of vision-based manipulation tasks from a large offline dataset of robot interaction. In order to accomplish this, humans need easy and effective ways of specifying tasks to the robot. Goal images are one popular form of task specification, as they are already grounded in the robot's observation space. However, goal images also have a number of drawbacks: they are inconvenient for humans to provide, they can over-specify the desired behavior leading to a sparse reward signal, or under-specify task information in the case of non-goal reaching tasks. Natural language provides a convenient and flexible alternative for task specification, but comes with the challenge of grounding language in the robot's observation space. To scalably learn this grounding we propose to leverage offline robot datasets (including highly sub-optimal, autonomously collected data) with crowd-sourced natural language labels. With this data, we learn a simple classifier which predicts if a change in state completes a language instruction. This provides a language-conditioned reward function that can then be used for offline multi-task RL. In our experiments, we find that on language-conditioned manipulation tasks our approach outperforms both goal-image specifications and language conditioned imitation techniques by more than 25%, and is able to perform visuomotor tasks from natural language, such as "open the right drawer" and "move the stapler", on a Franka Emika Panda robot.

cs.RO↗

arXiv2021

PMCE: efficient inference of expressive models of cancer evolution with high prognostic power

Motivation: Driver (epi)genomic alterations underlie the positive selection of cancer subpopulations, which promotes drug resistance and relapse. Even though substantial heterogeneity is witnessed in most cancer types, mutation accumulation patterns can be regularly found and can be exploited to reconstruct predictive models of cancer evolution. Yet, available methods cannot infer logical formulas connecting events to represent alternative evolutionary routes or convergent evolution. Results: We introduce PMCE, an expressive framework that leverages mutational profiles from cross-sectional sequencing data to infer probabilistic graphical models of cancer evolution including arbitrary logical formulas, and which outperforms the state-of-the-art in terms of accuracy and robustness to noise, on simulations. The application of PMCE to 7866 samples from the TCGA database allows us to identify a highly significant correlation between the predicted evolutionary paths and the overall survival in 7 tumor types, proving that our approach can effectively stratify cancer patients in reliable risk groups. Availability: PMCE is freely available at https://github.com/BIMIB-DISCo/PMCE, in addition to the code to replicate all the analyses presented in the manuscript. Contacts: daniele.ramazzotti@unimib.it, alex.graudenzi@ibfm.cnr.it.

stat.ML↗

arXiv2021

Janus solutions in three-dimensional N=8 gauged supergravity

Janus solutions are constructed in $d=3$, ${\cal N}=8$ gauged supergravity. We find explicit half-BPS solutions where two scalars in the $SO(8,1)/SO(8)$ coset have a nontrivial profile. These solutions correspond on the CFT side to an interface with a position-dependent expectation value for a relevant operator and a source which jumps across the interface for a marginal operator.

hep-th↗

arXiv2021

Investigation of the Stark Effect on a Centrosymmetric Quantum Emitter in Diamond

Quantum emitters in diamond are leading optically-accessible solid-state qubits. Among these, Group IV-vacancy defect centers have attracted great interest as coherent and stable optical interfaces to long-lived spin states. Theory indicates that their inversion symmetry provides first-order insensitivity to stray electric fields, a common limitation for optical coherence in any host material. Here we experimentally quantify this electric field dependence via an external electric field applied to individual tin-vacancy (SnV) centers in diamond. These measurements reveal that the permanent electric dipole moment and polarizability are at least four orders of magnitude smaller than for the diamond nitrogen vacancy (NV) centers, representing the first direct measurement of the inversion symmetry protection of a Group IV defect in diamond. Moreover, we show that by modulating the electric-field-induced dipole we can use the SnV as a nanoscale probe of local electric field noise, and we employ this technique to highlight the effect of spectral diffusion on the SnV.

quant-ph↗

arXiv2020

Topological Planning with Transformers for Vision-and-Language Navigation

Conventional approaches to vision-and-language navigation (VLN) are trained end-to-end but struggle to perform well in freely traversable environments. Inspired by the robotics community, we propose a modular approach to VLN using topological maps. Given a natural language instruction and topological map, our approach leverages attention mechanisms to predict a navigation plan in the map. The plan is then executed with low-level actions (e.g. forward, rotate) using a robust controller. Experiments show that our method outperforms previous end-to-end approaches, generates interpretable navigation plans, and exhibits intelligent behaviors such as backtracking.

cs.RO↗

arXiv2020

Learning Object-conditioned Exploration using Distributed Soft Actor Critic

Object navigation is defined as navigating to an object of a given label in a complex, unexplored environment. In its general form, this problem poses several challenges for Robotics: semantic exploration of unknown environments in search of an object and low-level control. In this work we study object-guided exploration and low-level control, and present an end-to-end trained navigation policy achieving a success rate of 0.68 and SPL of 0.58 on unseen, visually complex scans of real homes. We propose a highly scalable implementation of an off-policy Reinforcement Learning algorithm, distributed Soft Actor Critic, which allows the system to utilize 98M experience steps in 24 hours on 8 GPUs. Our system learns to control a differential drive mobile base in simulation from a stack of high dimensional observations commonly used on robotic platforms. The learned policy is capable of object-guided exploratory behaviors and low-level control learned from pure experiences in realistic environments.

cs.RO↗

arXiv2020

Holographic Line Defects in $D=4$, $N=2$ Gauged Supergravity

We construct solutions of four-dimensional $N=2$ gauged supergravity coupled to vector multiplets which are holographically dual to superconformal line defects. For the gauged STU and the $SU(1,n)/U(1)\times SU(n)$ coset models, we use the solutions to calculate holographic observables such as the expectation value of the defect and one-point functions in the presence of the defect.

hep-th↗

arXiv2020

Continuously tuning the refractive indices by restructuring materials on the 20-1000 atoms scale: improving anti-reflection coating designs

We demonstrate the capability of block-copolymer templating to tune the refractive indices of functional oxides over a broad range by structuring materials on the 20-1000 atoms scale, with simple one-pot synthesis. The presented method is then combined with genetic algorithm-based optimization to explore its application for anti-reflection coating design. Merging these techniques allows for the realization of a minimal two-layer anti-reflection stack for silicon with broadband reflectivity of just 3% from the nominal value of 38%, over a 120° angular span, validated by fabrication followed by optical measurements.

cond-mat.mes-hall↗