My views on anything

Comparing vision feature of Gemini and GPT

Comparing vision feature of Gemini and GPT

Comparing Vision Features of Gemini and GPT

In the field of computer vision and artificial intelligence, two models have been making waves: Gemini and GPT. The following article, https://arxiv.org/abs/2312.15011, compares these two models, their strengths, weaknesses, and implications for the broader field.

Table of Contents

Performance Comparison

Gemini and GPT have been evaluated on a range of vision tasks, including image generation, classification, segmentation, and captioning. The results of these evaluations, both quantitative and qualitative, are discussed in this section.

Strengths and Weaknesses

Each model has its own strengths and weaknesses. Gemini, with its transformer architecture and dual encoder-decoder, can leverage both global and local information, handle multimodal inputs and outputs, and generate diverse and coherent images. On the other hand, GPT, with its single autoregressive decoder, has its own set of advantages and disadvantages.

Novel Task: Image Editing

The paper proposes a novel task of image editing, where the model has to modify an existing image according to a natural language instruction. This task presents a new challenge for both models and opens up a new avenue for research.

Implications for Computer Vision and AI

The results of the comparison have far-reaching implications for the broader field of computer vision and artificial intelligence. These implications, as well as the potential directions for future research, are discussed in this section.

Reimagining leprosy elimination with AI analysis of a combination of skin lesion images with demographic and clinical data – The Lancet Regional Health – Americas

Reimagining leprosy elimination with AI analysis of a combination of skin lesion images with demographic and clinical data – The Lancet Regional Health – Americas

See the full paper here.

AI4Leprosy: A research project that aims to develop an AI-driven diagnosis assistant for leprosy, based on skin images and clinical data.

  • Dataset: The researchers collected 1229 skin images and 585 sets of metadata from 222 patients with leprosy or other dermatological conditions in a Brazilian leprosy referral center. The dataset is open-source and available for other researchers to use.
  • AI models: The researchers tested three AI models, using images and metadata both independently and in combination, to predict the probability of leprosy. They used convolutional neural networks (CNN) for image analysis and elastic-net logistic regression for metadata analysis.
  • Results: The best AI model achieved a high accuracy (90%) and area under curve (AUC) of 96.46% for leprosy diagnosis, using a combination of metadata and patient information. The most important clinical signs for leprosy were thermal sensitivity loss, nodules and papules, feet paresthesia, number of lesions and gender.
  • Implications: The AI model could be a useful tool to accelerate and improve leprosy diagnosis, especially in low-resource settings. The researchers plan to validate the model in larger and more diverse datasets, and to implement it in a smartphone app for frontline health workers.

Beyond building predictive models: TwinOps in biomanufacturing

Beyond building predictive models: TwinOps in biomanufacturing

On the wave of more and more manufacturers embracing the pervasive mission to build digital twins, also biopharmaceutical industry envisions a significant paradigm shift of digitalisation towards an intelligent factory where bioprocesses continuously learn from data to optimise and control productivity. While extensive efforts are made to build and combine the best mechanistic and data-driven models, there has not been a complete digital twin application in pharma. One of the main reasons is that production deployment becomes more complex regarding the possible impact such digital technologies could have on vaccine products and ultimately on patients. To address current technical challenges and fill regulatory gaps, this paper explores some best practices for TwinOps in biomanufacturing – from experiment to GxP validation – and discusses approaches to oversight and compliance that could work with these best practices towards building bioprocess digital twins at scale.

Please read our whole pre-print here: https://doi.org/10.36227/techrxiv.16478856.v1

Senior AI/ML engineer in Bengaluru, India at GSK

Senior AI/ML engineer in Bengaluru, India at GSK

I’m hiring a Senior AI/ML engineer in Bengaluru, India. You will work with the rest of our international team on delivering cutting edge AI/ML solutions to support our vaccines business. This is a great role to grow into a lead data scientist as well as developing your machine learning and modern DevOps skills.

https://gsk.wd5.myworkdayjobs.com/GSKCareers/job/India—Karnataka—Bengaluru/Senior-AIML-Engineer_272917

My next career step: GSK Vaccines

My next career step: GSK Vaccines

Weird day, after nearly 5yrs years at Microsoft I’ve handed in my badge and laptop. Very much excited about my next step that is even deeper into healthcare, but also sad leaving such a great company with amazing people behind. 

I cannot be more proud to join GSK as their new director of Analytics and AI. Their mantra feels like a homecoming: “We are a science-led global healthcare company with a special purpose: to help people do more, feel better, live longer.”