The global federated learning market revenue is expected to reach USD 260.5 million by 2030, exhibiting a CAGR of 10.7% during 2024–2030. This can be ascribed to the rapid acceptance of federated learning (FL) across a variety of sectors to improve learning about devices and organizations and its benefits over traditional learning such as data privacy. Moreover, it enables a large number of participants to create dependable deep learning models without actually sharing data from devices, ensuring data security, confidentiality, privileged access to information, and availability of large data sets for research.
One of the most innovative advancements in data science and artificial intelligence is federated learning. It could surpass the limitations of centralized machine learning. A group of decentralized edge devices or systems can train machine learning models using FL, also known as collaborative learning, without having to move or store the raw data on a central server. Training can be repeated on various servers or devices using FL applications.
Moreover, its operation is very easy to start and understand. A basic model is created by a central server and sent to edge machine learning operation (MLOp) devices. A pre-trained or untrained ML model is acquired by each device for use in subsequent training with the device's local data. The gadgets do not communicate with one another or the central server to exchange raw data. To the central server, only the biases, suggestions, and preferences data are sent. In order to create a collective global model, the server then adds the local models together.
In addition, model changes are examined to determine their accuracy and identify any advancements from the previous edition. Until the model achieves an acceptable degree of accuracy or until a predetermined number of iterations have been completed, the shared model technique is repeatedly applied.
The drug discovery category accounts for a significant share of the market. This is due to the increasing prevalence of diseases globally as a result of the growing population, surging pollution levels, and reducing the nutritious quality of food. This drives the need for effective drugs that are available or new drugs that are in the developing phase but have promising results. Drug discovery includes going through highly sensitive and valuable patient information, molecular structures, and experimental results.
Federated learning aids pharmaceutical companies and research institutions to come together on a single platform and train ML models on decentralized data. Each company collects and prepares its data relevant to drug discovery. This data consists of compound libraries, biological assays, clinical trial results, and scientific literature. It is used to train shared models, optimize model parameters using algorithms, and update models. After completion of local model training, the updated model parameters are integrated into one model. Then performance of that model is evaluated using metrics specific to drug discovery. Lastly, the model is deployed and integrated into drug discovery pipelines. Thus, these factors boost the market growth in this category.
Globally, the North American market is expected to witness considerable growth in the coming years. This can be attributed to the presence of well-known generated learning solution vendors, such as NVIDIA, IBM, Cloudera, Microsoft, and Google, in the region; and heavy investments in R&D specifically in IoT and AI to develop and improve FL techniques and algorithms.
The region has always been a global leader when it comes to technological advancements and innovation, making it a promising market. It has huge volumes of data from all sectors including healthcare, finance, retail, and manufacturing. This creates an ideal environment for advanced technologies like federated learning and lures industry players from all over the world.
Some of the key federated learning market players are NVIDIA, Cloudera, IBM, Microsoft, Google, Intel, Owkin, Intellegens, Edge Delta, and DataFleets.