Find the best Convolutional Neural Network jobs

Create CNN and Transformers models using the Pytorch framework. Data set is provided.

https://github.com/islamyasereq/LLM/tree/main?tab=readme-ov-file The paper only provided how to implement TrafficGPT with the repository but not for the CNNs. I just need small help with the CNNs so I can compare both for my project. I need assistance from an experienced data scientist or machine learning engineer to build and test Convolutional Neural Networks (CNNs) for encrypted traffic classification. The goal is to compare the CNN model's performance with the results of TrafficGPT (a GPT-2-based encrypted traffic classifier). Provided Resources: TrafficGPT repository with evaluation scripts. Preprocessed DC dataset. Dataset: The DC and USTC-TFC dataset (YouTube video traffic) is already preprocessed and ready to use. Requirements: Implement CNNs to classify the encrypted traffic. Use the CNN for feature extraction (exclude the softmax layer for feature embedding). Test the CNN using metrics like F1 score, closed-set accuracy, and open-set accuracy. Compare the CNN's performance to the provided TrafficGPT results. Expected Output: A functional CNN implementation with detailed performance metrics and comparison. I will provide all necessary resources upon hiring.

We are seeking an experienced developer to create a robust text detection solution for large diagrams. The ideal candidate will deliver near 100% accuracy in identifying text, whether horizontally or vertically aligned. The final output should consist of recognized text along with its coordinates. The project requires not only the implementation of the detection algorithm but also the provision of clean and well-documented source code. If you have a strong background in computer vision or machine learning, we would love to hear from you! Sample image is attached. Size can be slightly bigger or smaller. Disclaimer: Currently my algo is doing about 95% but I need more accuracy. The text is very clear. However the fonts can change from one image to another.

''YOLO, SSD, or Fast R-CNN'' Potentially using these elements, need to create a surveillance AI that detects when our patients and staff are not following the protocols, agreesive behavior, nurses needling their patients , need to spend 30 seconds cleaning the area first. etc. Looking for a solution that is already working that is open and we can implenent and further customize. Possibly with the following solutions: 1. **OpenPose**: - **Description**: An open-source library for real-time multi-person keypoint detection, including body, face, hands, and foot estimation. - **Application**: Can be used to monitor and analyze the movements of phlebotomists, ensuring they follow proper procedures like cleaning the arm for the required duration. - **Link**: OpenPose GitHub Repository 2. **DeepLabCut**: - **Description**: A toolbox for markerless pose estimation of animals and humans using deep learning. - **Application**: Suitable for tracking specific behaviors and postures, aiding in monitoring procedural adherence. - **Link**: DeepLabCut GitHub Repository 3. **OpenCV**: - **Description**: A comprehensive open-source computer vision and machine learning software library. - **Application**: Provides tools for image and video analysis, which can be employed to detect and analyze behaviors such as aggression among donors. - **Link**: OpenCV Official Website 4. **YOLO (You Only Look Once)**: - **Description**: A real-time object detection system. - **Application**: Can be trained to detect specific objects or behaviors within the video feed, such as identifying aggressive movements. - **Link**: YOLO GitHub Repository 5. **Surveillance Video Summarizer**: - **Description**: An AI-driven system that processes surveillance videos, extracts key frames, and generates detailed annotations. - **Application**: Assists in summarizing and reviewing surveillance footage to identify notable events and behaviors. - **Link**: Surveillance Video Summarizer GitHub Repository

We need an AI specialist to develop a tool capable of identifying and classifying patterns in medical images. The project requires expertise in deep learning, convolutional neural networks (CNNs), and frameworks like PyTorch or TensorFlow. Previous experience in medical AI projects is a plus.

Key Responsibilities: Develop, train, and optimize deep learning models for audio processing tasks such as speech recognition, speaker identification, noise suppression, and audio event detection. Research and implement state-of-the-art architectures and algorithms in audio deep learning, including self-supervised learning, generative models, and signal enhancement techniques. Collaborate with cross-functional teams to integrate audio AI models into Cisco’s products and services. Conduct rigorous testing, evaluation, and benchmarking of models against industry standards. Design efficient pipelines for real-time audio processing in edge and cloud environments. Prepare technical documentation and reports detailing methodologies, findings, and recommendations. Requirements: Ph.D. in Audio Deep Learning, Machine Learning, or a related field; OR a Master’s degree with 3+ years of professional experience in audio deep learning. Strong expertise in audio signal processing and machine learning techniques. Proficiency in deep learning frameworks such as TensorFlow, PyTorch, or Keras. In-depth understanding of advanced architectures like transformers, convolutional neural networks (CNNs), and recurrent neural networks (RNNs). Hands-on experience with audio datasets, preprocessing, feature extraction (e.g., MFCC, spectrograms), and augmentation. Strong programming skills in Python; experience with frameworks like Librosa, torchaudio, or similar is highly desirable. Knowledge of deploying deep learning models in production environments, including optimization for edge devices. Excellent problem-solving skills and ability to work independently. Nice-to-Have: Experience with speech-to-text systems, audio scene analysis, or related applications. Familiarity with distributed computing and model training on large-scale datasets. Background in telecommunications, networking, or similar industries. Published work in reputable journals or conferences in the field of audio deep learning. Why Work with Cisco? Be part of a global leader driving innovation in networking and communications. Work on cutting-edge projects with the potential to impact millions of users. Flexible work environment with the opportunity to collaborate with a world-class team. How to Apply: If you meet the requirements and are excited to tackle challenging problems in audio AI, please submit your application, including: A resume/CV highlighting your relevant experience. A brief cover letter detailing your expertise in audio deep learning. Links to any publications, projects, or GitHub repositories showcasing your work. We look forward to working with innovative minds who are ready to redefine the future of audio technology with us!

Description: I’m looking for a skilled programmer to develop an app that uses AI to solve medical survey questions with the intelligence of a logical, smart medical care specialist. The app should learn from data inputs over time and provide accurate, thoughtful answers to medical survey questions in line with professional standards. Key requirements include: AI Integration: Build a machine learning model tailored to medical decision-making. Survey Solution Logic: Train the AI to logically analyze and solve medical survey questions. User-Friendly Interface: Ensure the app is intuitive for users. Customization and Learning: Incorporate features for the AI to improve over time with user feedback. If you have experience in AI development, particularly in medical or logical applications, please share your portfolio and propose how you would approach this project. I'm open to suggestions to make this app the best it can be.

Intrusion Detection 1. Exploratory Data Analysis and data preprocessing • Understanding the dataset • Plot the distribution of categories to check whether data imbalance • Understand the properties of audio files 2. Feature Extraction 3. Model training and validation ● Train convolutional model with extracted features ● Train convolutional model with raw data 4. Testing the model and analyzing the model performance with different metrics and plots Performance evaluation ● Accuracy ● F1 score ● Recall Plots ● learning curve ● Performance metrics plots Tables ● Performance metrics tables 5. Compare the model with existing works and modification Documentation ● Results (2 pages) including figures and tables ● Discussion and Conclusion (2 pages) ● Comparison and discussion (1 page)

Project: Build a model to predict cryptocurrency behavior. Data: We provide a transformed dataset containing 240 features and a deviation variable. The data covers historical information from 2015 to March 2024 and focuses on the last four hours of cryptocurrency occurrences and observed behavior. (Note: The Google Drive link has been removed due to privacy concerns. We can share the data securely upon project selection.) Model Type: Multi-class classification model. Success Criteria: We're looking for a model with at least 90% accuracy measured by the combined metric of precision and recall. Compensation: We offer a $1,000 incentive for achieving the desired accuracy. Contact: Feel free to reach out with any questions or interest in the project Additional Information: We are open to discussing the possibility of providing additional data points if needed. Please let us know if you require further context about the project goals. Benefits of Working with Us: Contribute to an innovative project in the cryptocurrency space. Showcase your data science and machine learning skills. Earn a competitive reward for your expertise.

Overview We are looking for an experienced computer vision engineer to build an end-to-end system capable of processing images of individual items from a beverage fridge to predict the correct product accurately. Context We have a smart beverage refrigerator with 6 interior racks. When a rack is opened, the contents become visible to an array of cameras mounted on the top of the refrigerator. It uses a YOLO-based computer vision architecture to track objects and their location as they are added and removed from racks in a smart fridge using built-in cameras. This computer vision system generates undistorted, cropped images of the individual items in the fridge. This OBJECT DETECTION system has already been built. We are now looking to build the IMAGE RECOGNITION system. We are looking to build an additional computer vision pipeline capable of taking these individual images and predicting which product is in the image. We have a database of product data and accompanying product imagery from multiple angles to support this. System requirement - Predict the correct product with a high degree of accuracy. - Correctly identify the products in different lighting conditions. - Correctly identify the products when oriented at different angles. - Correctly identify the products when part of the object is occluded. - Handle frequent additions of new products. - Continuous learning from user feedback and new training data. - Flag low-confidence results for human review. - Bonus: connect off-the-shelf image recognition systems capable of performing the above actions versus building these in-house - Speed is not a criteria at this stage, 15 second processing times are fine. The candidate We are looking for a candidate who has delivered similar product recognition systems in the past, with experience in feature extraction, feature matching, similarity metrics, OCR, and other similar computer vision techniques. The candidate should be familiar with the ResNet family of models alongside alternatives such as CLIP, RegNet, and EfficientNet. Attached are several example images of objects from the refrigerator taken by the camera system and cropped by the existing computer vision system.