InfoQ Homepage Deep Learning Content on InfoQ
-
/articles/efficient-resource-management-small-language-models/en/smallimage/efficient-resource-management-logo-small-1730888361067.jpg)
Efficient Resource Management with Small Language Models (SLMs) in Edge Computing
Small Language Models (SLMs) bring AI inference to the edge without overwhelming the resource-constrained devices. In this article, author Suruchi Shah dives into how SLMs can be used in edge computing applications for learning and adapting to patterns in real-time, reducing the computational burden and making edge devices smarter.
-
/articles/pinterest-ad-ranking-ai/en/smallimage/UnpackinghowAds-1710929235364.jpg)
Unpacking How Ad Ranking Works at Pinterest
Aayush Mudgal describes how Pinterest serves advertisements. He discussed in detail how Machine Learning is used to serve ads at large scale. He went over ads marketplaces and the ad delivery funnel, the ad serving architecture, and two of the main problems: ad retrieval and ranking. Finally, he discussed some of the challenges and solutions for training and serving large models.
-
/articles/deep-learning-models-ai-interpretability-methods/en/smallimage/LOGO-1675867673135.jpg)
Understanding and Debugging Deep Learning Models: Exploring AI Interpretability Methods
ML interpretability refers to a user's ability to explain decisions made by an ML system. Interpretability increases confidence in the model, reduces bias, and ensures that model is compliant and ethical. In this article, author Andrew Hoblitzell discusses several methods of ML interpretability and dives deep into Local Interpretable Model-Agnostic Explanations (LIME) and Shapley Values.
-
/articles/building-neural-networks-tensorflow-net/en/smallimage/logo-1657201887957.jpg)
Building Neural Networks with TensorFlow.NET
TensorFlow is an open-source framework developed by Google scientists and engineers for numerical computing. TensorFlow.NET is a library that provides a .NET Standard binding for TensorFlow. In this article, the author explains how to use Tensorflow.NET to build a neural network.
-
/articles/deep-learning-institutional-incremental-learning/en/smallimage/institutional-incremental-learning-small-image-v2-1640874831613.jpg)
Developing Deep Learning Systems Using Institutional Incremental Learning
Institutional incremental learning promises to achieve collaborative learning. This form of learning can address data sharing and security issues, without bringing in the complexities of federated learning. This article talks about practical approaches which help in building an object detection system.
-
/articles/loosely-coupled-deep-learning-serving/en/smallimage/benefits-of-loosely-coupled-deep-learning-serving-small-image-1627300228992.jpg)
Benefits of Loosely Coupled Deep Learning Serving
As deep networks are becoming more specialized and resource-hungry, serving such networks on acceleration hardware in tight-budget environments is also becoming difficult. Instead of using API frameworks, loosely coupled components can be preferred as an alternative. They bring high controllability, easy adaptability, transparent observability, and cost-effectiveness when serving deep networks.
-
/articles/deep-learning-apache-spark-nvidia-gpu/en/smallimage/spark-logo-1623166971929.jpg)
Accelerating Deep Learning on the JVM with Apache Spark and NVIDIA GPUs
In this article, authors discuss how to use the combination of Deep Java Learning (DJL), Apache Spark v3, and NVIDIA GPU computing to simplify deep learning pipelines while improving performance and reducing costs. They also show the performance comparison of this solution with GPU vs CPU hardware, using Amazon EMR and NVIDIA RAPIDS Accelerator.
-
/articles/AI-Closer-Common-Sense/en/smallimage/AI-Closer-Common-Sense-s-1602793883610.jpeg)
Is Artificial Intelligence Closer to Common Sense?
Intelligent agents lack the common-sense knowledge they need to reason about the world. Traditionally, there have been two unsuccessful approaches to getting computers to reason about the world—symbolic logic and deep learning. A new project, called COMET, tries to bring these two approaches together. Although it has not yet succeeded, it offers the possibility of progress.
-
/articles/human-pose-estimation-ai-powered-fitness-apps/en/smallimage/human-pose-estimation-ai-powered-fitness-apps-s-1602698421552.jpeg)
Challenges of Human Pose Estimation in AI-Powered Fitness Apps
In this article, the author discusses the human pose estimation solution powered by AI technologies and the challenges faced in online fitness apps which use the pose estimation to predict the position of the human body based on an image or a video containing a person.
-
/articles/federated-machine-learning/en/smallimage/Figure-3-logo-1599498665208.jpeg)
Federated Machine Learning for Loan Risk Prediction
In this article, author Brendon Machado discusses how data owners and data scientists can work together to create models on privatized data using the federated learning technique and shows how to use it in loan risk prediction use cases.
-
/articles/gpt3-enabled-applications/en/smallimage/first-wave-GPT-3-enabled-applications-logo-small-1597055494711.jpg)
The First Wave of GPT-3 Enabled Applications Offer a Preview of Our AI Future
The first wave of GPT-3 powered applications are emerging. After priming of only a few examples, GPT-3 could write essays, answer questions, and even generate computer code! Furthermore, GPT-3 can perform algebraic calculations and language translations despite never being taught such concepts. However, GPT-3 is a black box with unpredictable outcomes. Developers must use it responsively.
-
/articles/java-machine-learning-djl/en/smallimage/java-logo-1590441719825.jpeg)
Machine Learning in Java with Amazon Deep Java Library
In this article, we demonstrate how Java developers can use the JSR-381 VisRec API to implement image classification or object detection with DJL’s pre-trained models in less than 10 lines of code.