Natural Language Processing advancements have escalated with the availability of powerful computing at lower costs, but training on GPUs is expensive. Lilt, an AI-powered enterprise translation software and services company, wanted to optimize training tasks using CPUs instead of GPUs. By optimizing TensorFlow on Intel Xeon 8380 processors, Lilt was able to increase increasing inference by nearly 4X and deploy their workloads on Google Cloud N2 high memory instances with Intel Optimizations
for TensorFlow 2.4.
Powered by Intel AI, Mindtree’s Cognitive Customer Service suite helps enterprises transform their omni-channel customer service via intelligent agent assistance, faster issue resolution, personalization and much more. This White paper shows how MindTree’s AI-powered implementation of contact centers using Intel Distribution of Python and SigOpt delivers amazing performance and productivity of the entire conversational AI pipeline.
Flutura 's PCB (printed circuit board) Defect Detection is a real-time system with automated failure analysis of PCB manufacturing. Flutura converted their PCB defect detection models for
OpenVINO and optimized them for parallel processing and post-quantization. They evaluated performance running on
2nd Gen Intel Xeon Scalable processor- and 3rd Gen Intel Xeon Scalable processor-based systems achieving significant performance improvement.
This brief highlights the EdgeVerve XtractEdge, a purpose-built document extraction, processing, and comprehension platform that unlocks business value from enterprise data by extracting intelligence from enterprise documents, regardless of complexity or domain specificity. The platform automates the end-to-end document processing lifecycle from ingestion to consumption, using AI capabilities such as Natural Language Processing (NLP), Computer Vision, and AI powered search utilizing Intel technologies to reduce time to value and operationalize models at enterprise scale.
This Solution Brief highlights how Intel optimizations of Winning Health’s Bone Age Assessment (BAA) model helped greatly reduce image analysis time enabling large scalability of SaaS solution for hospitals and clinicians on cloud platforms.
This Solution Brief highlights the Winning Health AI-powered medical imaging system, jointly optimized by Intel, Winning Health and AMAX (a global OEM.) The paper cites lung nodule detection as an example to demonstrate how Intel 2nd Generation Xeon processors coupled with Intel AI software e.g., DL Boost, OpenVINO, improve the inference performance in terms of speed and accuracy.
This brief discusses how the TietoEVRY edge reference platform is a viable alternative to existing commercial edge solutions to improve total cost of ownership (TCO) without compromising on performance. The platform, due its flexibility, scalability, cloud-native characteristic and Intel technology is powerful for emerging edge use cases across various industries such as Telecom, Industry, Enterprise, IoT, SmartCities, SmartHomes, Automotive or MedTech.
This brief discusses optimizing Matroid’s Similarity Search object detection model for Intel Xeon
Scalable processors has achieved increased performance and could open new doors for more flexible and possibly lower cost services and deployments.
This brief discusses Seassoon’s text detection solution which inferences data input for cognitive decision-making. It highlights how utilizing Intel Optimizations for PyTorch and Image
Detection enabled Seassoon to achieve 3X faster Inferencing and avoid the need for a more costly and complex GPU solution.
This brief discusses ICETech vision computing-based systems that automatically identify vehicles and license plates in unattended smart parking operations, allowing them to run more efficiently. Optimizing for OpenVINO and quantizing for INT8 using the Post Training Optimization toolkit (POT) and inferencing
with Intel DL Boost (VNNI) improved ICETech’s inferencing performance with minimal impact on accuracy.