Artificial Intelligence, Machine Learning, and Algorithmic Innovation
There is no doubt that today, some of the most important commercial technologies are based on mathematical or algorithmic innovation, particularly in the areas of Artificial Intelligence (AI) and Machine Learning (ML). O&R patent lawyers are uniquely positioned to provide unequaled representation to clients in these areas.
Our technical expertise in artificial intelligence spans decades, from early work in expert systems through recent work in knowledge representation and reasoning and natural language processing. O&R Patent Law combines tremendous technical background with practical experience navigating the pitfalls of the legal landscape to secure the intellectual property protection to which our clients are entitled.
AI and ML patents pose unique challenges, in part due to the evolving body of law and administrative procedures and the disparities between how such innovations are treated in different countries’ patent offices. A successful patent claim must avoid characterizing the innovation as an abstract concept, not patentable in the US, and at the same time have the requisite technical character required by certain foreign frameworks. We stay current on these issues in the US and significant foreign jurisdictions.
Machine learning
We have worked across the spectrum of machine learning techniques, from classical machine learning such as parametric modeling and optimization, through state-of-the-art modern techniques, such as Reinforcement Learning and Bayesian Optimization. Our Machine Learning background also includes design and use of Neural Network architectures, such as recurrent and deep architectures and transformer-based approaches. We find it important to put a client’s innovation in the context of this background in order to protect it successfully, for example, being able to distinguish between a new underlying technique, a new application, or a combination of the two.
Speech recognition and Natural Language Processing (NLP)
We have worked with speech recognition techniques as they have evolved from the use of template matching and Hidden Markov Models for automated transcription and word spotting to today’s approaches that use deep learning to map directly from input waveforms to text transcription or meaning. In natural language, we are familiar with classical computation linguistic techniques including grammar-based parsing and machine translation through today’s approaches for automated processing using machine learning techniques, including in the areas of information extraction, automated question answering, and transformer-based machine translation.
Our AI and ML experience also includes work in the following areas:
medical signal analysis
robot control
vehicle navigation
Mathematically based and Algorithmic Innovations
O&R can handle the complexities of today’s mathematically based and algorithmic innovations. Examples include audio processing improvements such as spatial audio, upmixing, or signal enhancement. Radio frequency techniques for system linearization or antenna control may at their heart be based on mathematical improvements or formulations. Cryptographic techniques, blockchain-based innovations, or new neural network approaches are often algorithmic. But today’s patent systems around the world have difficulty distinguishing such innovations from those that are “abstract” and not entitled to patent protection, no matter how new or non-obvious they are. This is familiar ground for our firm, and we can guide our clients through the sometimes challenging path to patent protection.