Personalized federated learning (PFL) has garnered significant attention for its ability to address heterogeneous client data distributions while preserving data privacy. However, when local client data is limited, deep learning models often suffer from insufficient training, leading to suboptimal performance. Foundation models, such as CLIP (Contrastive Language-Image Pretraining), exhibit strong feature extraction capabilities and can alleviate this issue by fine-tuning on limited local data. Despite their potential, foundation models are rarely utilized in federated learning scenarios, and challenges related to integrating new clients remain largely unresolved. To address these challenges, we propose the Dual Prompt Personalized Federated Learning (DP2FL) framework, which introduces dual prompts and an adaptive aggregation strategy. DP2FL combines global task awareness with local data-driven insights, enabling local models to achieve effective generalization while remaining adaptable to specific data distributions. Moreover, DP2FL introduces a global model that enables prediction on new data sources and seamlessly integrates newly added clients without requiring retraining. Experimental results in highly heterogeneous environments validate the effectiveness of DP2FL's prompt design and aggregation strategy, underscoring the advantages of prediction on novel data sources and demonstrating the seamless integration of new clients into the federated learning framework.

This work investigates the potential of Federated Learning (FL) for official statistics and shows how well the performance of FL models can keep up with centralized learning methods. FL is particularly interesting for official statistics because its utilization can safeguard the privacy of data holders, thus facilitating access to a broader range of data. By simulating three different use cases, important insights on the applicability of the technology are gained. The use cases are based on a medical insurance data set, a fine dust pollution data set and a mobile radio coverage data set - all of which are from domains close to official statistics. We provide a detailed analysis of the results, including a comparison of centralized and FL algorithm performances for each simulation. In all three use cases, we were able to train models via FL which reach a performance very close to the centralized model benchmarks. Our key observations and their implications for transferring the simulations into practice are summarized. We arrive at the conclusion that FL has the potential to emerge as a pivotal technology in future use cases of official statistics.

It seems that the world is gradually turning everything into data, and we are storing data at lightspeed. How are banks capitalising on new sources of data and using it to develop new services? Our watches now track our sleep, our movement, our heart rate and much more. With it they can "coach" us with training plans that are specific to our individual capabilities. At the top end of smartwatches, you have the ability to pay for things, listen to music, get guided GPS navigation and more.

Maor Shlomo is the CEO of external data platform, Explorium. When PageRank was developed in the 1990s, building powerful AI models was all about algorithms and architecture. In the 2000s, with the birth of Hadoop and large-scale data mining, we rushed to gather massive amounts of data in hopes of driving more accurate analytics. Finally, in the 2010s, cloud computing exploded, and the new infrastructure made training algorithms much easier and cheaper. All of these breakthrough technologies are now essential tools in the modern AI developer's sandbox.

Nearly three quarters of global financial services executives have admitted they are challenged by the fractured nature and vast amount of data available. The Aite Group surveyed 682 marketing and risk executives at financial institutions across five countries during the third quarter, finding that in the UK alone, 71 per cent of executives said they were challenged by the immense amount of data they have. The study found that the proliferation of artificial intelligence (AI) and machine learning (ML) is expected to continue over the next 24 months, with 68 per cent of UK executives - and three in four globally - considering integrating new analytics technology into their platforms. "Most financial institutions lack a single, cohesive analytics platform," said Tiffani Montez, senior analyst at Aite Group. "Firms may have vastly different data repositories and teams managing analytics functions, often leading to multiple approaches - by line of business, role and channel - across their institutions.

So, does AI really require a year of data preparation? The answer is profound: No. And clients typically see results for their first use case within 3-4 months. All business applications, including AI, used to be built based on the data model that governed a specific problem. Such a data model typically resides in a data warehouse that can be located either on-premise or in the cloud.

Subscribe to receive updates on Industry 4.0 The digital supply network (DSN) can be a powerful tool for companies, allowing them to harness data and information to make more effective decisions in the physical world via their assets, machines, and people. The benefits can be myriad: deeper visibility into the supply network; greater connectivity with suppliers, partners, and customers; smarter factories; and the ability to act, respond, and adapt intelligently to shifts in the ecosystem. Whatever the result, data--and the ability to analyze and derive insights from it--lies at the heart of the DSN. Companies may think they need to make significant infrastructure investments to realize these benefits, given the typical cost, complexity, and time to "rip and replace" existing applications. However, the road to a fully realized DSN does not necessarily equate to a wholesale replacement of IT assets. In fact, legacy systems are often able to support more DSN capabilities than previously imagined--and the data they already generate often contains significant potential.

The notion of using analytics to improve customer experience has changed the landscape and thought process of businesses over the past several years. As machine learning becomes further democratized, or more pervasively available, it is making its way into many enterprise software applications – including the Earnix software suite. Machine learning is impacting everyday analytical activities for our customers (segmentation, modeling, and optimization), and is improving very specific marketing program results. Often the ultimate goal of these marketing programs are to improve the customer experience – which requires a faster, more accurate, and more contextual customer interaction – which is better for both consumer and brand. I'll give a few examples of how machine learning is improving the customer experience through next best offer personalization, customer behavior analytics with new data sources, and analytical optimization.

In my first blog post on the topic – Customer Experience and Machine Learning: Practical Applications – I discussed how machine learning techniques are being used today by financial services organizations to achieve business benefit. Insurers and retail banks are using machine learning to improve personalization by being able to better analyze and predict customer behavior, and deliver the optimal marketing offer, message, or price. But what is coming in the future? Based on the research we are doing – we are seeing a few capabilities come to forefront. These include augmented analytics, collaborative machine learning, and the introduction of decision trees and neural networks within deep learning.

As machine learning algorithms are increasingly embedded in today's analytics applications, this offspring of artificial intelligence is grabbing headlines. In reality, however, the technology isn't as new as you might think. Machine learning algorithms have risen out of predictive analytics and use data to learn and improve behaviors and recognize patterns -- without being explicitly programmed to do so. While the technology may not be new, its business application are. To better understand the impact of machine learning, let's look how it intersects with customer experience and where this combination may be headed.