arXiv subjects

Puneet Jain

Publications and source records attributed to Puneet Jain.

11 recordsLinked to original sources

arXiv2026

KC-Agent: A Dual-Process Cognitive Architecture for Efficient ML Model Improvement

Data drift poses significant challenges for machine learning systems in production, requiring continuous model updates to maintain performance. We present KC-Agent, a dual-process cognitive architecture for automated ML model improvement that combines fast pattern recognition (System 1) with deliberate incremental updates (System 2). Our approach implements structured memory systems enabling System 1 to leverage successful solutions previously discovered by System 2, achieving efficient pattern-based responses without costly re-computation. KC-Agent incorporates atomic change principles and rollback capabilities to ensure reliable, verifiable updates in production environments. We evaluate our method on five datasets including real-world NASA turbofan data with authentic temporal degradation and synthetic datasets with controlled drift scenarios. KC-Agent achieves state-of-the-art performance (76.8% accuracy) while maintaining optimal efficiency (13.2s execution time), outperforming established cognitive architectures: CodeAct (+2.4%), Tree of Thoughts (+3.6%), ReAct (+8.0%), and Reflexion (+8.9%). Consensus evaluation by a panel of state-of-the-art LLMs confirms superior strategic efficacy (8.33/10 Smartness score), significantly outperforming baseline agents. The knowledge consolidation mechanism delivers 91% speedup over the slow variant while maintaining higher accuracy. Our approach demonstrates both theoretical foundations and practical viability for cognitive-inspired automated ML improvement systems capable of handling complex real-world data drift scenarios.

cs.AI↗

arXiv2026

Beyond Human Performance: A Vision-Language Multi-Agent Approach for Quality Control in Pharmaceutical Manufacturing

Colony-forming unit (CFU) detection is critical in pharmaceutical manufacturing, serving as a key component of Environmental Monitoring programs and ensuring compliance with stringent quality standards. Manual counting is labor-intensive and error-prone, while deep learning (DL) approaches, though accurate, remain vulnerable to sample quality variations and artifacts. Building on our earlier CNN-based framework (Beznik et al., 2020), we evaluated YOLOv5, YOLOv7, and YOLOv8 for CFU detection; however, these achieved only 97.08 percent accuracy, insufficient for pharmaceutical-grade requirements. A custom Detectron2 model trained on GSK's dataset of over 50,000 Petri dish images achieved 99 percent detection rate with 2 percent false positives and 0.6 percent false negatives. Despite high validation accuracy, Detectron2 performance degrades on outlier cases including contaminated plates, plastic artifacts, or poor optical clarity. To address this, we developed a multi-agent framework combining DL with vision-language models (VLMs). The VLM agent first classifies plates as valid or invalid. For valid samples, both DL and VLM agents independently estimate colony counts. When predictions align within 5 percent, results are automatically recorded in Postgres and SAP; otherwise, samples are routed for expert review. Expert feedback enables continuous retraining and self-improvement. Initial DL-based automation reduced human verification by 50 percent across vaccine manufacturing sites. With VLM integration, this increased to 85 percent, delivering significant operational savings. The proposed system provides a scalable, auditable, and regulation-ready solution for microbiological quality control, advancing automation in biopharmaceutical production.

cs.CV↗

arXiv2025

Challenging Disability and Interaction Norms in XR: Cooling Down the Empathy Machine in Waiting for Hands

Virtual Reality (VR) is often described as the "ultimate empathy machine," framing disability as an experience to be simulated through such technologies, which can reduce disability to a spectacle of pity or inspiration. In response, we present Waiting for Hands (WfH), an interactive eXtended Reality (XR) installation that critiques this logic by: (1) repurposing interaction norms in XR through the creation of Alternative Controllers, and (2) staging an absurd XR performance using the built controllers to disrupt sentimentalized disability narratives. The performance involves eight people: two XR participants on stage and six audience members watching a projected documentary about Hema Kumari, an Indian singer living with Rheumatoid Arthritis. The XR users partially obscure the film, drawing attention through strange mouth and hand movements performed in XR. This creates a layered experience that disrupts direct engagement with Hema's story and introduces uncertainty. While XR is often seen as a fully immersive, sensory-dominant medium, this piece subverts that framing by using XR to produce absurdity and alienation. By challenging empathy-driven and pitiable narratives of disability, we ask what ethical stance an XR performance can take to attune participants to non-normative embodiment while resisting spectacle.

cs.HC↗

arXiv2025

Assistive XR research for disability at ACM ASSETS: A Scoping Review

Despite the rise in affordable eXtended Reality (XR) technologies, accessibility still remains a key concern, often excluding people with disabilities from accessing these immersive XR platforms. Consequently, there has been a notable surge in HCI research on creating accessible XR solutions (also known as, assistive XR). This increased focus in assistive XR research is also reflected in the number of research and innovative solutions submitted at the ACM Conference on Accessible Computing (ASSETS), with an aim to make XR experiences inclusive for disabled communities. However, till date, there is little to no work that provides a comprehensive overview of state-of-the-art research in assistive XR for disability at ACM ASSETS, a premier conference dedicated for research in HCI for people with disabilities. This study aims to fill this research gap by conducting a scoping review of literature delineating the key focus areas, research methods, statistical and temporal trends in XR research for disability at ACM ASSETS (2019-2023). From a pool of 1595 articles submitted to ASSETS, 26 articles are identified that specifically focus on XR research for disability. Through a detailed analysis, 6 key focus areas of XR research explored at ACM ASSETS are identified and a detailed examination of each is provided. Additionally, an overview of multiple research methods employed for XR research at ASSETS is also presented. Lastly, this work reports on the statistics and temporal trends regarding the number of publications, XR technologies used, disabilities addressed, and methodologies adopted for assistive XR research at ASSETS, highlighting emerging trends and possible future research directions.

cs.HC↗

arXiv2024

Performance Prediction of Hub-Based Swarms

A hub-based colony consists of multiple agents who share a common nest site called the hub. Agents perform tasks away from the hub like foraging for food or gathering information about future nest sites. Modeling hub-based colonies is challenging because the size of the collective state space grows rapidly as the number of agents grows. This paper presents a graph-based representation of the colony that can be combined with graph-based encoders to create low-dimensional representations of collective state that can scale to many agents for a best-of-N colony problem. We demonstrate how the information in the low-dimensional embedding can be used with two experiments. First, we show how the information in the tensor can be used to cluster collective states by the probability of choosing the best site for a very small problem. Second, we show how structured collective trajectories emerge when a graph encoder is used to learn the low-dimensional embedding, and these trajectories have information that can be used to predict swarm performance.

cs.MA↗

arXiv2022

Error estimators and their analysis for CG, Bi-CG and GMRES

The demands of accuracy in measurements and engineering models today, renders the condition number of problems larger. While a corresponding increase in the precision of floating point numbers ensured a stable computing, the uncertainty in convergence when using residue as a stopping criterion has increased. We present an analysis of the uncertainty in convergence when using relative residue as a stopping criterion for iterative solution of linear systems, and the resulting over/under computation for a given tolerance in error. This shows that error estimation is significant for an efficient or accurate solution even when the condition number of the matrix is not large. An $\mathcal{O}(1)$ error estimator for iterations of the CG algorithm was proposed more than two decades ago. Recently, an $\mathcal{O}(k^2)$ error estimator was described for the GMRES algorithm which allows for non-symmetric linear systems as well, where $k$ is the iteration number. We suggest a minor modification in this GMRES error estimation for increased stability. In this work, we also propose an $\mathcal{O}(n)$ error estimator for A-norm and $l_{2}$ norm of the error vector in Bi-CG algorithm. The robust performance of these estimates as a stopping criterion results in increased savings and accuracy in computation, as condition number and size of problems increase.

math.NA↗

arXiv2022

Raveling the Role of Dopants on Charge Carrier Kinetics of TiO$_2$ Electrodes using Electrochemical Impedance Spectroscopy

We synthesized the pure and co-doped titanium dioxide (TiO$_2$) electrodes via spin coating. We examined the optical and electronic properties of as-prepared thin film electrodes with co-doping of transition metals and non-metals. The co-doping of Cu, Zn, and N increase the absorption of the radiation in the visible region. The doping leads to the formation of defect states in the electrodes. In this article, we have studied the carrier kinetics in pristine and co-doped TiO2 electrodes. To study the role of dopants in carrier transport of the synthesized TiO2 based electrodes, the electrochemical impedance spectroscopy measurement is performed in the frequency range of 10-1 Hz to 106 Hz. The study reveals the influence of dopants on electron-hole recombination in the defect sites present in bulk and the transport mechanism of the electrons and ions to the surface/interface of the electrodes.

physics.app-ph↗

arXiv2022

The Study of Hydrophilicity and Optical Properties of Zn and N-doped TiO$_{2}$ Thin Films

Among the wide bandgap semiconductors, TiO$_{2}$ is the highly stable and cost-efficient semiconductor used for the different photocatalysis processes like water splitting, chemical waste degradation, anti-micro bacterial activities, and more. Materials showing high hydrophilicity (surface phenomenon) with low bandgap are required to improve photocatalysis efficiency. We report the synthesis of Zn (4 wt.%) and N (4 wt.%) doped TiO$_{2}$ thin films using the spin coating technique to improve surface wettability. The XRD pattern shows the growth of the pure anatase phase of TiO$_{2}$. UV absorption spectra show a minor increment in the bandgap of the Zn and N doped TiO$_{2}$ thin films. The water contact angle with pure TiO$_{2}$ is 33.45° and reduces to 17.94° after 4wt% doping of Zn and N. The results show the enhanced hydrophilicity in the Zn and N doped TiO$_{2}$ thin films.

cond-mat.mtrl-sci↗

arXiv2020

Towards End-to-End In-Image Neural Machine Translation

In this paper, we offer a preliminary investigation into the task of in-image machine translation: transforming an image containing text in one language into an image containing the same text in another language. We propose an end-to-end neural model for this task inspired by recent approaches to neural machine translation, and demonstrate promising initial results based purely on pixel-level supervision. We then offer a quantitative and qualitative evaluation of our system outputs and discuss some common failure modes. Finally, we conclude with directions for future work.

cs.CL↗

arXiv2020

Dynamics of mosquito swarms over a moving marker

Insect swarms are a model system for understanding collective behavior where the collective motion appears in disorder. To initiate and maintain a swarm in place, flying insects often use a visual external cue called a marker. In mosquitoes, understanding the swarming behavior and its relation to the marker has an additional medical relevance since swarming often precedes mating in the wild, thus constituting an important stage to intercept for controlling mosquito population. In this paper, we conduct preliminary experiments to characterize the visual coupling between a swarm of mosquitoes and a marker. A laboratory microcosm with artificial lighting was built to stimulate consistent swarming in the malarial mosquito Anopheles stephensi. The experimental setup was used to film a mosquito swarm with a stereo camera system as a marker was moved back-and-forth with different frequencies. System identification analysis of the frequency response shows that the relationship between the swarm and the marker can be described by delayed second order dynamics in a feedback loop. Further, the length of the internal time delay appears to correlate with the number of mosquitoes swarming on the marker indicating that such a delay may be able capture social interactions within swarming systems. For insect swarms, model fitting of trajectory data provides a way to numerically compare swarming behaviors of different species with respect to marker characteristics. These preliminary results motivate investigating linear dynamic system in feedback as a framework for modeling insect swarms and set the stage for future studies.

physics.bio-ph↗

arXiv2014

Poster Abstract: Bits and Watts: Improving energy disaggregation performance using power line communication modems

Non-intrusive load monitoring (NILM) or energy disaggregation, aims to disaggregate a household's electricity consumption into constituent appliances. More than three decades of work in NILM has resulted in the development of several novel algorithmic approaches. However, despite these advancements, two core challenges still exist: i) disaggregating low power consumption appliances and ii) distinguishing between multiple instances of similar appliances. These challenges are becoming increasingly important due to an increasing number of appliances and increased usage of electronics in homes. Previous approaches have attempted to solve these problems using expensive hardware involving high sampling rates better suited to laboratory settings, or using additional number of sensors, limiting the ease of deployment. In this work, we explore using commercial-off-the-shelf (COTS) power line communication (PLC) modems as an inexpensive and easy to deploy alternative solution to these problems. We use the reduction in bandwidth between two PLC modems, caused due to the change in PLC modulation scheme when different appliances are operated as a signature for an appliance. Since the noise generated in the powerline is dependent both on type and location of an appliance, we believe that our technique based on PLC modems can be a promising addition for solving NILM.

cs.OH↗