Fine-tuning Research Articles

Precise interpretations of traditional fine-tuning measures

We uncover two precise interpretations of traditional electroweak fine-tuning (FT) measures that were historically missed. : the traditional FT measure shows the change in plausibility of a model in which a parameter was exchanged for the Z boson mass relative to an untuned model in light of the Z boson mass measurement. the traditional FT measure shows the exponential of the extra information, measured in nats, relative to an untuned model that you must supply about a parameter in order to fit the Z mass. We derive the mathematical results underlying these interpretations, and explain them using examples from weak scale supersymmetry. These new interpretations allow us to rigorously define FT in particle physics and beyond, shed fresh light on the status of extensions to the Standard Model and, lastly, allow us to precisely reinterpret historical and recent studies using traditional FT measures. Published by the American Physical Society 2025

Influence of Process Parameter and Build Rate Variations on Defect Formation in Laser Powder Bed Fusion SS316L.

Laser powder bed fusion (LPBF) is an additive manufacturing process that has gained interest for its material fabrication due to multiple advantages, such as the ability to print parts with small feature sizes, good mechanical properties, reduced material waste, etc. However, variations in the key process parameters in LPBF may result in the instantiation of porosity defects and variation in build rate. Particularly, volumetric energy density (VED) is a variable that encapsulates a number of those parameters and represents the amount of energy input from the laser source to the feedstock. VED has been traditionally used to inform the quality of the printed part but different values of VED are presented as optimal values for certain material systems. An optimal VED value can be maintained by changing the key process parameters so that various combinations yield a constant value. In this study, an optimal constant VED value is maintained while printing SS316L with variable key processing parameters. Porosity analysis is performed using optical microscopy, as well as X-ray computed tomography, to reveal the volume density and distribution of those pores. Two primary defect categories are identified, namely lack of fusion and porosity induced by balling defects. The findings indicate that, even at optimal VED, variations in process parameters can significantly influence defect type, underscoring the sensitivity of defect formation to the variation of these parameters. Furthermore, a minor change in the build rate, driven by adjustments in process parameters, was found to influence defect categories. These findings emphasize that fine tuning the process parameters and build rate is essential to minimize defects. Finally, fiducial marks have been identified as a source of unintentional porosity defects. These results enable the refinement of process parameters, ultimately optimizing LPBF to achieve enhanced material density and expedite the printing.

Effect of gelatin/xanthan gum ratios on jackfruit leaf extract nanoemulsion gel stability and properties

Nanoemulsions were formulated as hydrogels, incorporating extract from jackfruit leaf (JLE) and consisting of a blend of xanthan gum and fish gelatin (XAN-GEL). Utilizing the spontaneous emulsification technique, single xanthan gum (XAN), single fish gelatin (GEL), and XAN-GEL blends were hydrated at a mass ratio (0.5 to 2.5% w/v) and were prepared and subsequently assessed for various parameters such as droplet size, polydispersity index (PDI), zeta potential, colloidal stability, and viscosity. Additionally, plain nanoemulsion (NE) and carbomer were synthesized as control samples for comparative analysis. The droplet size, PDI, and zeta potential values of the developed nanoemulsion and gelled nanoemulsion fell within the ranges of 131 to 223 nm, 0.23 to 0.33, and -27.3 to -47.3 mV, respectively. Notably, all formulations exhibited stability except for the lower amount of XAN ratio blends at 1.0% (MD) and 0.5% (ME). Furthermore, all stable nanoemulsion gels demonstrated shear thinning behavior, and the highest amount of XAN ratio blends at 2.0% (MA) enhanced the viscosity of the nanoemulsion by fine tuning the rheological characteristics of the targeted gelled nanoemulsion suitable for future topical drug delivery application.

Pulsed-Current Operation Enhances H2O2 Production on a Boron-Doped Diamond Mesh Anode in a Zero-Gap PEM Electrolyzer.

A niobium (Nb) mesh electrode was coated with boron-doped diamond (BDD) using chemical vapor deposition in a custom-built hot-filament reactor. The BDD-functionalized mesh was tested in a zero-gap electrolysis configuration and evaluated for the anodic formation of H2O2 by selective oxidation of water, including the analysis of the effects on Faradaic efficiency towards H2O2 (FEH2O2) induced by pulsed electrolysis. A low electrolyte flow rate (V⋅anolyte) was found to result in a relatively high concentration of H2O2 in single-pass electrolysis experiments. Regarding pulsed electrolysis, we show an optimal ratio of on-time to off-time to obtain the highest concentration of H2O2. Off-times that are "too short" result in decreased FEH2O2, whereas "too long" off-times dilute the product in the electrolyte stream. Using our electrolyzer setup with an anodic pulse of 2 s with 4 s intervals, and a V⋅anolyte of 0.75 cm3 min-1, resulted in the best performance. This adjustment increased the FEH2O2 by 70 % compared to constant current electrolysis, at industrially relevant current densities (150 mA cm-2). Fine tuning of BDD morphology, flow patterns, and anolyte composition might further increase the performance of zero-gap electrolyzers in pulsed operation modes.

Fine tuning enzyme activity assays for monitoring the enzymatic hydrolysis of PET

Efficient monitoring of the enzymatic PET-hydrolysis is crucial for developing novel plastic-degrading biocatalysts. Herein, we aimed to upgrade in terms of accuracy the analytical methods useful for monitoring enzymatic PET-degradation. For the HPLC-based assessment, the incorporation of an internal standard within the analytic procedure enabled a more accurate quantification of the overall TPA content and the assessment of molar distributions and relative content of each aromatic degradation product. The provided calibration curves cover a broad concentration range, from µM to low mM scale, facilitating assessment of both lower and higher PETase activities, with a limit of detection positioned below the reported PET-degrading activities. The increased reproducibility and accuracy of the improved HPLC method, compared to the previous methods, was supported by lower dispersion of product concentrations and their lower deviation from theoretical values over multiple measurements. The other predominantly employed UV-spectroscopy assay was also improved in terms of employed wavelength and medium extinction coefficient of the three aromatic degradation products, while being cross-validated by the improved HPLC method. Finally, both methods were used for monitoring the product formation within the leaf-branch compost cutinase (LCC)-mediated PET-hydrolysis and provided individual time-productivity profiles for each aromatic degradation product.

Fine Tuning of an Advanced Planner for Cognitive Training of Older Adults.

Developing effective cognitive training tools for older adults, specifically addressing executive functions such as planning, is a challenging task. It is of paramount importance to ensure the implementation of engaging activities that must be tailored to the specific needs and expectations of older adults. Furthermore, it is essential to provide the appropriate level of complexity for the planning task. A human-centred approach was used to address the issues identified in the design of the tool. Two pilot studies were conducted with older adults to fine-tune the training task and optimize its suitability for them. This also led to an enhancement of the underlying planning engine, transitioning from a simple fast-forward planner (PDDL4J) to an advanced heuristic search planner (ENHSP). The results show that user studies enabled the development of a cognitive training system that gradually increased the proposed difficulty levels of the planning task while maintaining usability and satisfaction among older adults. This highlights the importance of conducting user studies when implementing cognitive training tools for older adults.

Adsorption-induced segregation as a way to control the catalytic performance of palladium-based bimetallic catalysts

The review analyzes recent publications devoted to the experimental and computational studies of adsorption-induced segregation effects in palladium-based bimetallic catalysts. The segregation processes are considered for two types of systems, those based on substitutional solid solutions (alloys) and on intermetallic compounds. The applicability of adsorptioninduced segregation effects for fine tuning of active sites on the catalytic surface is discussed. The possibility of controlling catalytic properties in various reactions using these effects is analyzed. The prospects for the development of this research area are given in the conclusion. <br> The bibliography includes 138 references.

Neutrophil behavior during the metabolic adaptations to short term high fat feeding

Background and Aims: Neutrophils participate to the chronic metabolic consequences of High Fat Diet (HFD). Nevertheless, neutrophils are characterized by short half-life which is determined by a fine tuning in expression and function of CXCR4, which dictates the egress from the bone marrow (BM), and CXCR2, which facilitates their mobilization from BM and the patrolling activity in the periphery. In the quest to study the behavior of the neutrophil with the short-term consequences of HFD feeding, we studied whether the metabolic adaptations affect blood neutrophil count and the membrane expression of their indirect markers of function. Methods: To assess the gluco-metabolic impact of a short-term HFD feeding, indirect calorimetry and plasma glucose dosage were performed on mice previously fed a HFD (60% Kcal from fat) for seven days, followed by immunophenotyping of blood over 24 hours. To test whether changes in circulating neutrophil count during short-term HFD feeding were related to a differential egress from the BM, we repeated the same experimental design in mice harboring a conditional deletion of CXCR4 (CXCR4fl/flMrp8Cre+). Results: Short-term HFD feeding was sufficient to induce a profound metabolic impact (e.g. reduced respiratory exchange ratio, increased energy expenditure, and insulin levels), and to induce an increase of circulating neutrophils (p=0,052), without impacting other leukocytic fractions over 24 hours, compared to chow diet feeding mice (20% Kcal from fat). HFD feeding significantly altered the expression pattern of multiple membrane markers of neutrophil function (CD11b, CD62L, CXCR2) over 24 hours, driving neutrophils toward a phenotype featuring increased migration and activation. Finally, the CXCR4fl/flMrp8Cre+ mice, which present significantly higher circulating neutrophilia, showed lower insulin sensitivity upon HFD compared to WT. Conclusions: We suggest that the metabolic adaptations induced by a short-term exposure to HFD affect neutrophil behavior, surmising it as an appealing target for cardio-metabolic diseases.

Viral coat proteins decrease the gene silencing activity of cognate and heterologous viral suppressors

Plant viruses have evolved different viral suppressors of RNA silencing (VSRs) to counteract RNA silencing which is a small RNA-mediated sequence-specific RNA degradation mechanism. Previous studies have already shown that the coat protein (CP) of cucumber mosaic virus (CMV) reduced RNA silencing suppression (RSS) activity of the VSR of CMV, the 2b protein. To demonstrate the universality of this CP-VSR interference, our study included three different viruses: CMV and peanut stunt virus (PSV) from the Bromoviridae, and plum pox virus (PPV) from the Potyviridae family. The RSS activity of the three VSRs (CMV 2b, PSV 2b, and PPV HC-Pro) was compared using Agrobacterium-mediated transient expression in Nicotiana benthamiana and the effect of CMV CP, PSV CP and PPV CP was validated on the RSS activity of their cognate and heterologous VSRs as well. Furthermore, the VSRs were also evaluated in PTGS suppressor-deficient CMV mutant (CMV NVE/10–12/AAA) virus-infected plants. The joint presence of CPs and VSRs resulted in decreased RSS activity in each combination, regardless of the origin of the two proteins, suggesting a universal role of the viral CPs in fine tuning of RSS. Interestingly the PSV CP elicited the strongest negative effect on the RSS activity of all three VSRs.

Transfer learning with inception V4 for behavioral anomaly detection in CCTV surveillance

Heightened security concerns and better affordability have driven the proliferation of video surveillance systems. Developing advanced CCTV cameras capable of real-time analysis of behavioral patterns to find abnormal patterns is highly anticipated and will help avert hazardous incidents. Deep Learning approaches, specifically Convolutional Neural Network (CNN) architecture-based approaches, have been extensively used for anomaly detection tasks in video analytics. The study results from research applying diversified Inception V4 transfer learning methodologies to accurately and efficiently identify abnormal activities in video surveillance systems. This study utilized the Inception V4 framework to classify video keyframes that are representative of normal or abnormal behavior. This paper will elaborate on applying transfer learning techniques of pre-training and fine-tuning, which will employ the required attributes from the input information to build a specialized predictor. The effectiveness of the presented models was evaluated through experimental studies that utilized UCF-Crime data for training and testing. Metrics, such as accuracy, recall, precision, and F1 scores, were employed as evaluation criteria to assess the performance of each model. The Fine-Tuned (F-T) model demonstrated performance metrics of 930%, 91.74%, 88.33%, and 90.01%, whereas the Pre-trained (P-T) model showed 88.70%, 88.93%, 87.12%, and 88.02%, respectively. These findings suggest that Transfer Learning (TL), employing the Inception V4 architecture, can effectively distinguish between normal and abnormal video behaviors. Moreover, adjusting the weights of particular layers through fine-tuning can improve the performance.

Electrochemical Synthesis and Conductivity Fine Tuning of the 2D Iron-Quinoid Metal-Organic Framework.

Electrically conducting 2D metal-organic frameworks (MOFs) with hexagonal 2D lattices like other 2D van der Waals stacked materials are attracting increasing interest. The conductivity can be effectively regulated through electronic structure adjustment thanks to the chemical and physical flexibility and adjustability of MOFs. In this regard, through a simple and rapid electrochemical method, 2D conductive iron-quinoid MOFs were synthesized. The conductivity of the obtained MOF film reached 1.7(7) S/m. With the increase of reaction time, the 2D network was oxidized partially, and the conductivity decreased down to 0.5(7) S/m. The DFT calculation results showed a narrow bandgap of the 2D crystal cell. Further quantum chemical calculations of the bimetal unit of the iron-quinoid MOF revealed the expansion of the bandgap as the 2D MOF network is gradually oxidized. This work proves the feasibility of fine-tuning macroscopic conductivity from an electronic structure. The combined organic and inorganic chemical structure of MOF materials provides a larger operating space for this fine regulation than that of pure inorganic materials. This conductive 2D iron-quinoid MOF film provides a new option for the development of novel microelectronic devices.

Application of large language models in clinical record correction: a comprehensive study on various retraining methods.

We evaluate the effectiveness of large language models (LLMs), specifically GPT-based (GPT-3.5 and GPT-4) and Llama-2 models (13B and 7B architectures), in autonomously assessing clinical records (CRs) to enhance medical education and diagnostic skills. Various techniques, including prompt engineering, fine-tuning (FT), and low-rank adaptation (LoRA), were implemented and compared on Llama-2 7B. These methods were assessed using prompts in both English and Spanish to determine their adaptability to different languages. Performance was benchmarked against GPT-3.5, GPT-4, and Llama-2 13B. GPT-based models, particularly GPT-4, demonstrated promising performance closely aligned with specialist evaluations. Application of FT on Llama-2 7B improved text comprehension in Spanish, equating its performance to that of Llama-2 13B with English prompts. Low-rank adaptation significantly enhanced performance, surpassing GPT-3.5 results when combined with FT. This indicates LoRA's effectiveness in adapting open-source models for specific tasks. While GPT-4 showed superior performance, FT and LoRA on Llama-2 7B proved crucial in improving language comprehension and task-specific accuracy. Identified limitations highlight the need for further research. This study underscores the potential of LLMs in medical education, providing an innovative, effective approach to CR correction. Low-rank adaptation emerged as the most effective technique, enabling open-source models to perform on par with proprietary models. Future research should focus on overcoming current limitations to further improve model performance.

Optimal pre-train/fine-tune strategies for accurate material property predictions

A pathway to overcome limited data availability in materials science is to use the framework of transfer learning, where a pre-trained (PT) machine learning model (on a larger dataset) can be fine-tuned (FT) on a target (smaller) dataset. We systematically explore the effectiveness of various PT/FT strategies to learn and predict material properties and create generalizable models by PT on multiple properties (MPT) simultaneously. Specifically, we leverage graph neural networks (GNNs) to PT/FT on seven diverse curated materials datasets, with sizes ranging from 941 to 132,752. Besides identifying optimal PT/FT strategies and hyperparameters, we find our pair-wise PT-FT models to consistently outperform models trained from scratch on target datasets. Importantly, our MPT models outperform pair-wise models on several datasets and, more significantly, on a 2D material band gap dataset that is completely out-of-domain. Finally, we expect our PT/FT and MPT frameworks to accelerate materials design and discovery for various applications.

Thermodynamical topology with multiple defect curves for dyonic AdS black holes

Dyonic black holes with quasitopological electromagnetism exhibit an intriguing phase diagram with two separated first-order coexistence curves. In this paper, we aim to uncover its influence on the black hole thermodynamical topology. At first, we investigate the phase transition and phase diagram of the dyonic black holes. Comparing with previous study that there is no black hole phase transition region for a middle pressure, we find this region can narrow or disappear by fine tuning the coupling parameter. Instead, two first-order phase transitions can be observed. Importantly, we uncover that such novel phase diagram shall lead to a multiple defect curve phenomenon in black hole topology where each dyonic black hole is treated as one defect in the thermodynamical parameter space. By examining the topology, it is shown that there could be one, three, or five black hole states for given pressure and temperature. For each case, the topological number is calculated. Our results show that the topological number always takes value of +1, keeping unchanged even when the multiple defect curves appear. Therefore, our study provides an important ingredient on understanding the black hole thermodynamical topology.

Development and Evaluation of Machine Learning Models for the Detection of Emergency Department Patients with Opioid Misuse from Clinical Notes.

The accurate identification of Emergency Department (ED) encounters involving opioid misuse is critical for health services, research, and surveillance. We sought to develop natural language processing (NLP)-based models for the detection of ED encounters involving opioid misuse. A sample of ED encounters enriched for opioid misuse was manually annotated and clinical notes extracted. We evaluated classic machine learning (ML) methods, fine-tuning of publicly available pretrained language models, and a previously developed convolutional neural network opioid classifier for use on hospitalized patients (SMART-AI). Performance was compared to ICD-10-CM codes. Both raw text and text transformed to the United Medical Language System were evaluated. Face validity was evaluated by term feature importance. There were 1123 encounters used for training, validation, and testing. Of the classic ML methods, XGBoost had the highest AU_PRC (0.936), accuracy (0.887), and F1 score (0.863) which outperformed ICD-10-CM codes [accuracy 0.870; F1 0.830]. Logistic regression, support vector machine, and XGBoost models had higher AU_PRC using transformed text, while decision trees performed better using raw text. Excluding XGBoost, fine-tuned pre-trained language models outperformed classic ML methods. The best performing model was the fine-tuned SMART-AI based model with domain adaptation [AU_PRC 0.948; accuracy 0.882; F1 0.851]. Explainability analyses showed the most predictive terms were 'heroin', 'opioids', 'alcoholic intoxication, chronic', 'cocaine', 'opiates', and 'suboxone'. NLP-based models outperform entry of ICD-10-CM diagnosis codes for the detection of ED encounters with opioid misuse. Fine tuning with domain adaptation for pre-trained language models resulted in improved performance.

Automated Detection of Spine Deformities: Advancing Orthopedic Care with Convolutional Neural Networks

This paper proposes Spine-CNN, a deep learning model for the detection of spinal deformities that can assist orthopedic doctors as a reliable tool for diagnosis. This technology promises to dramatically simplify the diagnostic process, freeing valuable time, and resources for healthcare professionals. To achieve this objective, a dataset of spine deformity X-ray images was curated from the PhysioNet database. The Spine-CNN was specially designed for detecting the spine deformity by incorporating features to leverage its ability to extract intricate features from radiographic images and by fine tuning the hyperparameters to properly train the model. Model performance was evaluated using standard metrics. Results from the Spine-CNN demonstrated promising performance in detecting spinal deformities. The model achieved an accuracy of 74%, with precision, recall, and F1-score values of 77%, 70%, and 73% respectively. Specifically, this research work introduces a Spine-CNN that underscore the potential of deep learning techniques to revolutionize diagnostic practices in orthopedic medicine, leading to improved treatment outcomes and patient care. Keywords: Computer-aided detection, Convolutional neural network, Image classification, Spine Deformation, X-ray imaging

Quantum Wire Coupled to Light

Experimental advances in cavity QED are raising the prospect of using light to probe quantum materials beyond the linear response regime. The capability to access quantum coherent phenomena would significantly advance the field. However, theoretical work on many-body systems coupled to light in the quantum coherent regime has been select. Here, we investigate the radiative properties of a finite-sized quantum wire in a microwave cavity. Examples of quantum wires include single-walled carbon nanotubes, a key experimental system in the field of nano-optics and plasmonics. We find that, for a variety of excited states, the repeated emission of photons results in the generation of many-body quantum entanglement. This leads to an increase in the rate at which subsequent photons are emitted, an example of Dicke superradiance. On the other hand, Pauli blocking tends to reduce this effect. Bosonization, the description of the excitations of a one-dimensional electron system as a gas of bosons, is found to be a powerful theoretical tool in this context. Its application means that many of our results generalize to wires with strong electron-electron interactions. The quantum wire thus represents a new platform to realize Dicke-model physics that does not rely on the various fine tunings necessary in traditional realizations involving many spatially isolated emitters. More broadly, this work demonstrates how quantum entanglement can be generated and measured in a many-body system. Published by the American Physical Society 2024

Advancements in Arabic Sign Language Recognition: A Method based on Deep Learning to Improve Communication Access

This research anticipates the issue of real-time sign language recognition as a means of facilitating the daily communication of the hearing-impaired persons. Improvement in the rate of real time sign language recognition will help in boosting interaction in the society and access to services for the hearing-impaired persons. This paper outlines a step-by-step guide on fine tuning an object detection model for the AASL using the SSD architecture. The AASL dataset which includes total of 7,857 samples of more than 200 patients are first preprocessed where the images are scaled, normalized, and augmented. In SSD model, the base of the network is the VGG16 network and few extra layers for feature extraction and for auxiliary and prediction of objects are added. Particularly in object detection, Intersection over Union (IoU), and mean Average Precision (mAP) are used in addition to the confusion matrix metrics comprising accuracy, precision, recall, and F1-score. The proposed model provides good recognition accuracy of 98 percent. This feature checks its capability of real-time identification of sign languages and shows 25% efficiency. The efficiency of the proposed method is explained when comparing the results of the modified VGG16-based SSD with other methodologies. As for the improvement of communication for the hearing-impaired individuals, the present work demonstrates that the deep learning methods can be significantly effective; however, suggesting that more research efforts should be directed to real-time solutions and that the datasets should be expanded.

Structural basis of orientated asymmetry in a mGlu heterodimer

The structural basis for the allosteric interactions within G protein-coupled receptors (GPCRs) heterodimers remains largely unknown. The metabotropic glutamate (mGlu) receptors are complex dimeric GPCRs important for the fine tuning of many synapses. Heterodimeric mGlu receptors with specific allosteric properties have been identified in the brain. Here we report four cryo-electron microscopy structures of mGlu2-4 heterodimer in different states: an inactive state bound to antagonists, two intermediate states bound to either mGlu2 or mGlu4 agonist only and an active state bound to both glutamate and a mGlu4 positive allosteric modulator (PAM) in complex with Gi protein. In addition to revealing a unique PAM binding pocket among mGlu receptors, our data bring important information for the asymmetric activation of mGlu heterodimers. First, we show that agonist binding to a single subunit in the extracellular domain is not sufficient to stabilize an active dimer conformation. Single-molecule FRET data show that the monoliganded mGlu2-4 can be found in both intermediate states and an active one. Second, we provide a detailed view of the asymmetric interface in seven-transmembrane (7TM) domains and identified key residues within the mGlu2 7TM that limits its activation leaving mGlu4 as the only subunit activating G proteins.

Fine tuning the large language pegasus model for dialogue summarization

Fine tuning the large language pegasus model for dialogue summarization