Search results
Ericsson training to suit your learning style and competence needs. New technology implementations or digital transformations don’t just end with deployment. Building employee competence to maximize the benefits of technologies and processes is crucial.
- Exploring machine learning use cases in telecom - Ericsson
Machine learning use cases: how to design ML architectures...
- Exploring machine learning use cases in telecom - Ericsson
- Machine Learning System Components
- Data Collection and Monitoring Components
- Model Registry
- Model Serving
- Offline vs. Online Predictions
- Workflow Management
- CI/CD, System Monitoring and Control
- Summary
- Acknowledgements
- Learn More
Machine learning models exist within a complex ecosystem and depend on numerous services to function properly (see Figure 2). ML models may ingest data from simple user inputs, distributed databases, or streaming data pipelines. Some models have minimal storage requirements, but others depend on large, ML optimized data storage. Across the end-to-e...
Data is the lifeline of any ML system. Telecom data is complex, multi-modal, and plentiful, comprising numerical metrics and text-based logs collected from many thousands of devices. The computational and communication costs of processing the data, as well as the latency and performance requirements, determines how the data components should be des...
Model registries are used to track the specific versions of model weights, parameters and code during the course of a ML system’s lifecycle. During deployment, models may be retrained on newer data, hyperparameters may be optimized based on new metrics and code may be updated to improve performance. Careful tracking and logging assists in identifyi...
Model serving frameworks exist to reliably provide responses to prediction requests. As part of this work, they provide a unified and predictable interface that abstracts away the details of a model’s underlying inference pipeline. Popular open-source frameworks for model serving include Seldon Core and KF Serving, which operate within the microser...
Machine learning predictions can be made in either periodically scheduled batches (offline), or in a dynamic streaming manner in real time (online). Batch prediction may be suitable when some delay is acceptable. In batch prediction, model prediction requests are accumulated over time, and the model responds to each batch of requests at an appropri...
To orchestrate the entire end-to-end ML pipeline, workflow management tools can help immensely. An ML pipeline consists of a number of inter-dependent tasks including data collection, transformation, validation, training, and serving. Workflow management tools can help effectively chain these tasks together, such that unexpected delays or issues in...
Since an ML solution is a deployed set of services that run together as a system, ML teams will benefit from adopting system administration best practices. These could include techniques such as CI/CD, monitoring, visualization and alerting. Like any software project, an ML project can use a Continuous Integration (CI) pipeline to run a set of test...
ML solutions are complex systems comprised of several components that may differ from the existing infrastructure organizations have in place. Depending on the particular use case, each of these sub-components may be implemented in a different manner. In future blog posts we will delve into more details about these use cases, and how they take can ...
In writing this post, I would like to thank Xuancheng Fan, Kunal Rajan Deshmukh, David Stone, and Zhaoji Huang for their contributions in related work, and Zeljka Lemaster for proof reading and revision suggestions.
How to build robust anomaly detectors with machine learning. Here’san introduction to machine reasoning in networks. Read more about AI in telecom networks. Read more about CI/CD.
Search through the complete Ericsson technology training portfolio and access course descriptions. Contact learning services. Get in touch with our offices in the US, Latin America, Europe, Middle-East, Asia and Oceania. Frequently asked questions. Find answers to your questions about Ericsson training portfolio, sales and delivery. Latest news
Jan 1, 2016 · In this interview conducted by McKinsey’s Simon London, Bina Chaurasia, Ericsson’s chief human-resources officer, describes how the company has revamped HR in response—increasing its agility, coordination, global scale, and ability to leverage data analytics.
On this site you will gain access to Ericsson’s curated learning content with focus on modern technologies. You will learn amazing things about how fields such as Artificial Intelligence are drastically changing our world today.
Oct 15, 2021 · In her interview with McKinsey’s Asin Tavakoli, Sonia shares an overview of Ericsson’s new data operating model and discusses the importance of focusing on people to deliver this change, including clearly articulating the vision to all stakeholders, empowering employees to take ownership of data, and providing company-wide data education.
Jan 11, 2021 · Vidya Krishnan, chief learning officer at Ericsson, outlines the most critical components of a successful learning and development culture.
