Choosing a compression technique from the data's characteristics is abstract when the claim never says how

2023-08-02Fed. Cir.

Realtime Data LLC v. Array Networks Inc., Nos. 2021-2251, 2021-2291 (Fed. Cir. Aug. 2, 2023) (nonprecedential) (Reyna, J., joined by Taranto, J.; Newman, J., dissenting), affirming dismissals by the United States District Court for the District of Delaware in Nos. 1:17-cv-00800-CFC and 1:17-cv-00925-CFC (Connolly, C.J.)

Realtime Data LLC asserted United States Patent Nos. 9,054,728, 8,933,825, 8,717,203, 9,116,908, 7,415,530, 10,019,458 and 9,667,751. Claim 25 of the '728 patent recites "analyzing, using a processor, data within a data block to identify one or more parameters or attributes" and choosing between content dependent compression and a single encoder. Claim 1 of the '908 patent recites a data accelerator that compresses two blocks with different techniques so that "the compression and storage occurs faster than the first and second data blocks are able to be stored on the memory device in uncompressed form." After a 2020 remand for a fuller analysis, Chief Judge Connolly held the claims ineligible, allowed amendment, and dismissed again. The panel majority, Judge Reyna joined by Judge Taranto, affirmed. The majority agreed with the district court that the claims are directed to "manipulating information using compression," and "[b]ecause data compression is, without more, simply a form of data analysis, the claims are directed to abstract ideas." "[N]either the claim nor the specification ever explains how that data is to be analyzed or compressed," nor "how the storage of the compressed data occurs 'faster.'" Unlike the patent in Visual Memory, "The patents here, by contrast, fail to explain the 'how.'" Judge Newman dissented: "This is properly an enablement case," and Section 101 "was never intended to bar categories of invention in this way."

What it changesSelecting an algorithm from the characteristics of the data is the core of every adaptive pipeline in machine learning, from choosing an encoder by content type to routing a query to the model best suited to it, and this panel held it abstract when the claim recites the selection as a result and the specification describes the candidate techniques only by name. The claim should recite the specific analysis, which attributes are measured and how, and the specific selection mechanism, and the specification should show the technical gain with numbers rather than asserting that storage occurs faster. The decision is nonprecedential and the dissent would have sent the case to Section 112, but the majority's demand for the "how" matches the court's precedential line from Electric Power Group forward, and a compression or routing portfolio should be reviewed against it before it is asserted. Seven patents fell on the pleadings, without claim construction or discovery.

More on Eligibility of software and AI claims

← All recent developments · Search all developments

Burbage Law is the practice of Jon Burbage, a registered patent attorney. 342 United States patents have issued from applications he wrote; the full list is on the Patents page. More about Jon Burbage.

This is general information about a published decision, agency guidance or piece of legislation. It is not legal advice and does not create an attorney-client relationship. Items concerning offices outside the United States are reported for information only and are not counsel on the law of those countries.