Privacy Enhancement for Cloud-Based Few-Shot Learning

Parnami, Archit; Usama, Muhammad; Fan, Liyue; Lee, Minwoo

doi:10.1109/ijcnn55064.2022.9892135

Cited by 10 publications

(10 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With a sufficiently diverse set of pictures to use as training data that include a variety of poses, lighting conditions, and backgrounds, a model trained using the methodology described herein is able to re‐identify even unseen individuals without the need for re‐training the model as seen in the open‐set evaluation metrics. This is achievable due to the inherent FSL properties (Parnami & Lee, 2022) of the methodology and will be especially useful in re‐identification of free‐ranging animals.…”

Section: Discussionmentioning

confidence: 99%

“…One of the key developments in machine learning that made FSL methods attractive for the individual re‐identification problem domain was the development of loss functions specifically designed to work with small amounts of data (Parnami & Lee, 2022), such as a loss function known as triplet loss. A loss function is a metric that measures the error between the model's prediction and the ground truth data.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An open‐source general purpose machine learning framework for individual animal re‐identification using few‐shot learning

Wahltinez,

Wahltinez

2024

Methods Ecol Evol

View full text Add to dashboard Cite

Animal re‐identification remains a challenging problem due to the cost of tagging systems and the difficulty of permanently attaching a physical marker to some animals, such as sea stars. Due to these challenges, photo identification is a good fit to solve this problem whether evaluated by humans or through machine learning. Accurate machine learning methods are an improvement over manual identification as they are capable of evaluating a large number of images automatically and recent advances have reduced the need for large training datasets. This study aimed to create an accurate, robust, general purpose machine learning framework for individual animal re‐identification using images both from publicly available data as well as two groups of sea stars of different species under human care. Open‐source code was provided to accelerate work in this space. Images of two species of sea star (Asterias rubens and Anthenea australiae) were taken using a consumer‐grade smartphone camera and used as original datasets to train a machine learning model to re‐identify an individual animal using few examples. The model's performance was evaluated on these original sea star datasets which contained between 39–54 individuals and 983–1204 images, as well as using six publicly available re‐identification datasets for tigers, beef cattle noses, chimpanzee faces, zebras, giraffes and ringed seals ranging between 45–2056 individuals and 829–6770 images. Using time aware‐splits, which are a data splitting technique ensuring that the model only sees an individual's images from a previous collection event during training to avoid information leaking, the model achieved high (>99%) individual re‐identification mean average precision for the top prediction (mAP@1) for the two species of sea stars. The re‐identification mAP@1 for the mammalian datasets was more variable, ranging from 83% to >99%. However, this model outperformed published state‐of‐the‐art re‐identification results for the publicly available datasets. The reported approach for animal re‐identification is generalizable, with the same machine learning framework achieving good performance in two distinct species of sea stars with different physical attributes, as well as seven different mammalian species. This demonstrates that this methodology can be applied to nearly any species where individual re‐identification is required. This study presents a precise, practical, non‐invasive approach to animal re‐identification using only basic image collection methods.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

An open‐source general purpose machine learning framework for individual animal re‐identification using few‐shot learning

Wahltinez,

Wahltinez

2024

Methods Ecol Evol

View full text Add to dashboard Cite

show abstract

“…As the workflow shown in Figure 1, then, an approach using the few‐shot learning (FSL) method (FSL Abs ) was built to decipher the relationship from Abs value to the relative intensity of molecular markers in 12 samples determined by the ESI‐FT‐ICR‐MS (see Text S6 in Supporting Information ) (Wright & Ziegler, 2017). FSL is an algorithm of the ML that is aimed to learn the underlying pattern from a few samples (Parnami & Lee, 2022), and has been widely used in previous object detection (Kisantal et al., 2019), cheminformatics (Chen et al., 2023), and environmental studies (Huang et al., 2023). Here, we used the synthetic minority over‐sampling technique (SMOTE) (Chawla et al., 2002) combined with the random forest (RF) algorithm provided by Ranger package (Fan et al., 2023; Hong, Cao, Fan, Lin, Bao, et al., 2022; Wright & Ziegler, 2017) to successfully build an FSL model without the risk of overfitting (has been proved in a 55 × 35799 ESI‐FT‐ICR‐MS data set, see details in Text S8 of the Supporting Information ) (Belgiu & Drăguţ, 2016; Cortes‐Ciriano & Bender, 2015; Jablonka et al., 2020), and then proved by the validation data set and model outputs (Text S9 in Supporting Information ) (Arulkumaran et al., 2017).…”

Section: Methods and Data Analysismentioning

confidence: 99%

Nitrogen‐Containing Functional Groups Dominate the Molecular Absorption of Water‐Soluble Humic‐Like Substances in Air From Nanjing, China Revealed by the Machine Learning Combined FT‐ICR‐MS Technique

Hong,

Zhang,

Bao

et al. 2023

JGR Atmospheres

View full text Add to dashboard Cite

The light absorption capacity of water‐soluble humic‐like substances (HULISWS) at the molecular level is crucial for reducing the uncertainties in modeling the radiative forcing. This study proposed a machine learning approach to allocate the light absorption coefficient at 365 nm (Abs365) of HULISWS into 8084 Fourier transform‐ion cyclotron resonance mass spectrometry (FT‐ICR‐MS) detached molecular markers and their potential functional groups. The ML model showed an acceptable uncertainty (<5%) to the whole Abs365 value based on the prediction errors. The results showed that five critical light‐absorbing molecules (C4H6O4NS, C8H6O4NS, C11H15O3N2, C12H15O3N2, and C19H21O6) could explain 74% (±3%) of the variation of Abs365 in the winter, whereas no crucial light‐absorbing molecules were found in the summer. Besides, the nitrogen‐containing functional groups were found to dominate (61% ± 8%) the molecular absorption near the 365 nm of the spectrum. This work illustrated how functional groups affect the absorption of HULISWS, providing critical information for future research of HULISWS on the molecular level.

show abstract

“…Fine-tune the original LLM model, which is computationally expensive; 11,12 2. Prompting within the LLM, which only accommodates small amounts of data and requires iterative user input; 13–15 and 3. Retrieval-augmented generation (RAG).…”

Section: Introductionmentioning

confidence: 99%

Development of a Liver Disease-Specific Large Language Model Chat Interface using Retrieval Augmented Generation

Ge,

Sun,

Owens

et al. 2023

Preprint

View full text Add to dashboard Cite

BackgroundLarge language models (LLMs) have significant capabilities in clinical information processing tasks. Commercially available LLMs, however, are not optimized for clinical uses and are prone to generating incorrect or hallucinatory information. Retrieval-augmented generation (RAG) is an enterprise architecture that allows embedding of customized data into LLMs. This approach “specializes” the LLMs and is thought to reduce hallucinations.MethodsWe developed “LiVersa,” a liver disease-specific LLM, by using our institution’s protected health information (PHI)-complaint text embedding and LLM platform, “Versa.” We conducted RAG on 30 publicly available American Association for the Study of Liver Diseases (AASLD) guidelines and guidance documents to be incorporated into LiVersa. We evaluated LiVersa’s performance by comparing its responses versus those of trainees from a previously published knowledge assessment study regarding hepatitis B (HBV) treatment and hepatocellular carcinoma (HCC) surveillance.ResultsLiVersa answered all 10 questions correctly when forced to provide a “yes” or “no” answer. Full detailed responses with justifications and rationales, however, were not completely correct for three of the questions.DiscussionsIn this study, we demonstrated the ability to build disease-specific and PHI-compliant LLMs using RAG. While our LLM, LiVersa, demonstrated more specificity in answering questions related to clinical hepatology – there were some knowledge deficiencies due to limitations set by the number and types of documents used for RAG. The LiVersa prototype, however, is a proof of concept for utilizing RAG to customize LLMs for clinical uses and a potential strategy to realize personalized medicine in the future.

show abstract

Privacy Enhancement for Cloud-Based Few-Shot Learning

Cited by 10 publications

References 18 publications

An open‐source general purpose machine learning framework for individual animal re‐identification using few‐shot learning

An open‐source general purpose machine learning framework for individual animal re‐identification using few‐shot learning

Nitrogen‐Containing Functional Groups Dominate the Molecular Absorption of Water‐Soluble Humic‐Like Substances in Air From Nanjing, China Revealed by the Machine Learning Combined FT‐ICR‐MS Technique

Development of a Liver Disease-Specific Large Language Model Chat Interface using Retrieval Augmented Generation

Contact Info

Product

Resources

About