Protecting 'Smart Algorithms' Under Patent Law

If you ask a patent law expert what you should do if certain novel subject matters fall under patent law exclusions that do not allow you to obtain patent protection, the first reaction from the expert would be to keep it a trade secret like a formula of Coca-Cola or Google algorithms.

Most of the secrecy managers in any company routinely check how intellectual property like software, engineering designs, and marketing plans are available in digital database or document form; they say you can have protected them as an intellectual property right. But when they are asked how to protect an algorithm that is not a document or database?

The answer of the legal expert varies from country to country as the national laws relating to patenting algorithms are not the same in all jurisdictions. For example, Section 3(k) bars patenting algorithms in India, which reads "mathematical methods, business methods, computer programs per se, and algorithms" under the nonpatentable inventions category under Section 3. The CRI guidelines 2017 also clarify what should and should not be allowed with respect to mathematical methods, business methods, computer programs per se, or algorithms.

Position in India After Telefonaktiebolaget L.M. Ericsson case

In an appeal against refusing a patent for "A Chaos Theoretical Exponent Value Calculation system" [3624/DELNP/2005] where the Deputy Controller held that the functions of the so-called system are based on a mathematical method for solving mathematical equations and declined to accept the technical effect theory followed under European Patent law, the Controller ruled that Indian patent law does not allow a patent for mathematical methods which have a technical effect. Therefore, the invention was rejected as not patentable under S.3(k) of the Patents Act 1970. Affirming the order of the Controller, Intellectual Property Appellate Board in Telefonaktiebolaget L.M. Ericsson vs Controller of Patents (2014) 08 IPAB CK 0006 held that

Similarly, the Deutsche Telekom Ag application (1960/DEL/1997) for "Procedure and device used to load input data during authentication into an algorithm" and Daewoo Electronics Co application (1602/CAL/1997) for the "Method and apparatus for contour motion estimating a binary image by using a weighted block match algorithm" were refused by the patent office. Bharatplanet.Com.Ltd application (291/MAS/2000) for "A method, system and algorithm for dynamic transliteration from the English language to any other language" also met a similar fate. No appeal was filed in these cases.

These decisions of the patent office left the seekers of patents on algorithms high and dry. Consequently, in the early years of the implementation of Section 3(k), some of the applicants preferred to either withdraw applications relating to algorithms or allow them to be abandoned under Section 21(1). For example, Thomson Licensing S.A withdrew their application (2942/DELNP/2004) for "A Memory Management Algorithm for Trellis Decoders". Porayath Rajiv also opted out, and his application (1140/MUM/2000) for "Acoustic feedback elimination using adaptive notch filter algorithm" was treated as abandoned under Section 21(1).

Similarly, Qualcomm's two applications (IN/PCT/2002/381/CHE and IN/PCT/2002/382/CHE) for "Modified finger assignment algorithm for high data rate calls" were deemed to abandoned under Section 21(1). Koninklijke Philips Electronics N.V application (IN/PCT/2002/626/CHE) for "Method of running an algorithm and a scalable programmable processing device also met the same fate, and so did the fate of Indian Institute of Technology application (690/DEL/2004) for "Architecture of a rearrangeable nonblocking all-optical cross-connect & the associated algorithm thereof".

Seeing the fate of algorithm-related applications at the patent office and the reasons for the exception of algorithms under Section 3(k), patent law experts rightly insisted that companies cannot patent an algorithm as such. This belief was affirmed on the premise that traditionally algorithms are simply telling a computer what to do. Section 3(k) specifically bars patents on algorithms. This means no patent protection is available for algorithms as such.

Patenting Novel Algorithms

Earlier case law, even in the US, categorically suggested algorithms were considered as an abstract idea and not eligible for patents. This position changed when the patent for the application of algorithms in the chemical industry was considered patentable. Later, a patent for the revolutionary "PageRank algorithm" for Google's online search was granted as it was for a specific method that produced a new and useful result. The position of patenting algorithms in the US now is guided by the Supreme Court case, Alice Corp. vs. CLS Bank International, which ruled that abstract ideas implemented using a computer were not eligible for a patent unless the patent adds something extra that embodies an innovative concept. This means if an algorithm is used in a specific and novel way that results in a technical effect, it may cross the patentability hurdle.

Novel Proprietary Algorithms Patentable: Examples

Proprietary algorithms offer new or significantly improved techniques for solving problems or enhancing the company's existing processes. Most talked about and popular patented algorithms are listed below:

Indian Patents on algorithms related inventions

If we see the records of granted Indian patents, we will find that Twentynine patents have been granted for inventions relating to algorithms so far, and this list is growing. The list of granted patents is given below:

Conclusion

These granted patents in India and elsewhere clearly show that novel smart algorithms have also impacted various non-traditional, non-software fields. Innovative companies are racing to develop the next game-changing smart algorithm, machine learning tool, AI bot, computer vision technology, and a lot more. Smart algorithms are replacing traditional methods of R&D search enquiry ranging from marketing and sales, manufacturing, supply chain planning, medicinal chemistry, computational chemistry and so on. In the new digital world, we will find the use of a trained neural network to control the fabrication of chemicals like rubber, glass, a detergent or the controlled development of a biological or biomedical compound or drug, a vaccine, an artificial limb, an artificial organ, lab-grown meat etc.

In the rapidly evolving landscape of the use of digital technology and innovation, smart algorithms have emerged as powerful tools that drive artificial intelligence (AI) and machine learning applications. Some smart novel algorithms form the company's valuable assets and can produce a competitive advantage. This means proprietary algorithms that are unique, not easily replicable, and capable of leveraging work processes to improve efficiency, drive innovation, and create new products or services with added competitive advantages in the market. The current legal landscape for patenting algorithms in India, like the USA, is complex and somewhat uncertain. This uncertainty primarily arises from the IPAB decision discussed above.

However, the patent examiners in India are looking west, particularly the ruling of Alice Corp. vs. CLS Bank International, and this has made it significantly more challenging to patent algorithms, as many may be viewed as mere abstract ideas. The list of granted patents above shows that the position of patenting algorithms in India is also rosy. An expert help can guide how an algorithm, when used in a specific, novel way to solve a technical problem, rather than the algorithm itself, can cross the bar placed under Section 3(k). Mind you, the innovative smart algorithm should be claimed as part of a system or process; it must clearly demonstrate its functionality, novelty, and non-obviousness to cross the non-patentability cliff.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.