AI and machine learning have something of an image problem.

Theyve never been quite so widely discussed as topics, or, arguably, their potential so widely debated. This is, to some extent, part of the problem. Artificial Intelligence can, still, be anything, achieve anything. But until its results are put into practice for people, it remains a misunderstood concept, especially to the layperson.

While well-established industry thought leaders are rightly championing the fact that AI has the potential to be transformative and capable of a wide range of solutions, the lack of context for most people is fuelling fears that it is simply going to replace peoples roles and take over tasks, wholesale. It also ignores the fact that AI applications have been quietly assisting peoples jobs, in a light touch manner, for some time now and people are still in those roles.

Many people are imagining AI to be something it is not. Given the technology is still in a fast-development phase, some people think it is helpful to consider the tech as a type of plug and play, black box technology. Some believe this helps people to put it into the context of how it will work and what it will deliver for businesses. In our opinion, this limits a true understanding of its potential and what it could be delivering for companies day in, day out.

The hyperbole is also not helping. The statements we use AI and our products AI driven have already become well-worn by enthusiastic salespeople and marketeers. While theres a great sales case to be made by that exciting assertion, its rarely speaking the truth about the situation. What is really meant by the current use of artificial intelligence? Arguably, AI is not yet a thing in its own right; i.e the capability of machines to be able to do the things which people do instinctively, which machines instinctively do not. Instead of being excited by hearing the phrase we do AI!, people should see it as a red flag to dig deeper into the technology and the AI capability in question.

SEE ALSO:

Machine learning, similarly, doesnt benefit from sci-fi associations or big sales patter bravado. In its simplest form, while machine learning sounds like a defined and independent process, it is actually a technique to deliver AI functions. Its maths, essentially, applied alongside data, processing power and technology to deliver an AI capability. Machine learning models dont execute actions or do anything themselves, unless people put them to use. They are still human tools, to be deployed by someone to undertake a specific action.

The tools and models are only as good as the human knowledge and skills programming them. People, especially in the legal sectors autologyx works with, are smart, adaptable and vastly knowledgeable. They can quickly shift from one case to another, and have their own methods and processes of approaching problem solving in the workplace. Where AI is coming in to lift the load is on lengthy, detailed, and highly repetitive tasks such as contract renewals. Humans can get understandably bored when reviewing highly repetitive, vast volumes of contracts to change just a few clauses and update the document. A machine learning solution does notnget bored, and performs consistently with a high degree of accuracy, freeing those legal teams up to work on more interesting, varied, or complicated casework.

Together, AI, machine learning and automation are the arms and armour businesses across a range of sectors need to acquire to adapt and continue to compete in the future. The future of the legal industry, for instance, is still a human one where knowledge of people will continue to be an asset. AI in that sector is more focused on codifying and leveraging that intelligence and while the machine and AI models learn and grow from people, so those people will continue to grow and expand their knowledge within the sector too. Today, AI and ML technologies are only as good as the people power programming them.

As Oren Etzioni, CEO of the Allen Institute for Artificial Intelligence put it, AI is neither good nor evil. Its a tool. A technology for us to use. How we choose to apply it is entirely up to us.

By Ben Stoneham, founder and CEO, autologyx

Follow this link:
Removing the robot factor from AI - Gigabit Magazine - Technology News, Magazine and Website

In a major operators network control center complaints are flooding in. The network is down across a large US city; calls are getting dropped and critical infrastructure is slow to respond. Pulling up the systems event history, the manager sees that new 5G towers were installed in the affected area today.

Did installing those towers cause the outage, or was it merely a coincidence? In circumstances such as these, being able to answer this question accurately is crucial for Ericsson.

Most machine learning-based data science focuses on predicting outcomes, not understanding causality. However, some of the biggest names in the field agree its important to start incorporating causality into our AI and machine learning systems.

Yoshua Bengio, one of the worlds most highly recognized AI experts, explained in a recent Wired interview: Its a big thing to integrate [causality] into AI. Current approaches to machine learning assume that the trained AI system will be applied on the same kind of data as the training data. In real life it is often not the case.

Yann LeCun, a recent Turing Award winner, shares the same view, tweeting: Lots of people in ML/DL [deep learning] know that causal inference is an important way to improve generalization.

Causal inference and machine learning can address one of the biggest problems facing machine learning today that a lot of real-world data is not generated in the same way as the data that we use to train AI models. This means that machine learning models often arent robust enough to handle changes in the input data type, and cant always generalize well. By contrast, causal inference explicitly overcomes this problem by considering what might have happened when faced with a lack of information. Ultimately, this means we can utilize causal inference to make our ML models more robust and generalizable.

When humans rationalize the world, we often think in terms of cause and effect if we understand why something happened, we can change our behavior to improve future outcomes. Causal inference is a statistical tool that enables our AI and machine learning algorithms to reason in similar ways.

Lets say were looking at data from a network of servers. Were interested in understanding how changes in our network settings affect latency, so we use causal inference to proactively choose our settings based on this knowledge.

The gold standard for inferring causal effects is randomized controlled trials (RCTs) or A/B tests. In RCTs, we can split a population of individuals into two groups: treatment and control, administering treatment to one group and nothing (or a placebo) to the other and measuring the outcome of both groups. Assuming that the treatment and control groups arent too dissimilar, we can infer whether the treatment was effective based on the difference in outcome between the two groups.

However, we can't always run such experiments. Flooding half of our servers with lots of requests might be a great way to find out how response time is affected, but if theyre mission-critical servers, we cant go around performing DDOS attacks on them. Instead, we rely on observational datastudying the differences between servers that naturally get a lot of requests and those with very few requests.

There are many ways of answering this question. One of the most popular approaches is Judea Pearl's technique for using to statistics to make causal inferences. In this approach, wed take a model or graph that includes measurable variables that can affect one another, as shown below.

To use this graph, we must assume the Causal Markov Condition. Formally, it says that subject to the set of all its direct causes, a node is independent of all the variables which are not direct causes or direct effects of that node. Simply put, it is the assumption that this graph captures all the real relationships between the variables.

Another popular method for inferring causes from observational data is Donald Rubin's potential outcomes framework. This method does not explicitly rely on a causal graph, but still assumes a lot about the data, for example, that there are no additional causes besides the ones we are considering.

For simplicity, our data contains three variables: a treatment , an outcome , and a covariate . We want to know if having a high number of server requests affects the response time of a server.

In our example, the number of server requests is determined by the memory value: a higher memory usage means the server is less likely to get fed requests. More precisely, the probability of having a high number of requests is equal to 1 minus the memory value (i.e. P(x=1)=1-z , where P(x=1) is the probability that x is equal to 1). The response time of our system is determined by the equation (or hypothetical model):

y=1x+5z+

Where is the error, that is, the deviation from the expected value of given values of and depends on other factors not included in the model. Our goal is to understand the effect of on via observations of the memory value, number of requests, and response times of a number of servers with no access to this equation.

There are two possible assignments (treatment and control) and an outcome. Given a random group of subjects and a treatment, each subject has a pair of potential outcomes: and , the outcomes Y_i (0) and Y_i (1) under control and treatment respectively. However, only one outcome is observed for each subject, the outcome under the actual treatment received: Y_i=xY_i (1)+(1-x)Y_i (0). The opposite potential outcome is unobserved for each subject and is therefore referred to as a counterfactual.

For each subject, the effect of treatment is defined to be Y_i (1)-Y_i (0) . The average treatment effect (ATE) is defined as the average difference in outcomes between the treatment and control groups:

E[Y_i (1)-Y_i (0)]

Here, denotes an expectation over values of Y_i (1)-Y_i (0)for each subject , which is the average value across all subjects. In our network example, a correct estimate of the average treatment effect would lead us to the coefficient in front of x in equation (1) .

If we try to estimate this by directly subtracting the average response time of servers with x=0 from the average response time of our hypothetical servers with x=1, we get an estimate of the ATE as 0.177 . This happens because our treatment and control groups are not inherently directly comparable. In an RTC, we know that the two groups are similar because we chose them ourselves. When we have only observational data, the other variables (such as the memory value in our case) may affect whether or not one unit is placed in the treatment or control group. We need to account for this difference in the memory value between the treatment and control groups before estimating the ATE.

One way to correct this bias is to compare individual units in the treatment and control groups with similar covariates. In other words, we want to match subjects that are equally likely to receive treatment.

The propensity score ei for subject is defined as:

e_i=P(x=1z=z_i ),z_i[0,1]

or the probability that x is equal to 1the unit receives treatmentgiven that we know its covariate is equal to the value z_i. Creating matches based on the probability that a subject will receive treatment is called propensity score matching. To find the propensity score of a subject, we need to predict how likely the subject is to receive treatment based on their covariates.

The most common way to calculate propensity scores is through logistic regression:

Now that we have calculated propensity scores for each subject, we can do basic matching on the propensity score and calculate the ATE exactly as before. Running propensity score matching on the example network data gets us an estimate of 1.008 !

We were interested in understanding the causal effect of binary treatment x variable on outcome y . If we find that the ATE is positive, this means an increase in x results in an increase in y. Similarly, a negative ATE says that an increase in x will result in a decrease in y .

This could help us understand the root cause of an issue or build more robust machine learning models. Causal inference gives us tools to understand what it means for some variables to affect others. In the future, we could use causal inference models to address a wider scope of problems both in and out of telecommunications so that our models of the world become more intelligent.

Special thanks to the other team members of GAIA working on causality analysis: Wenting Sun, Nikita Butakov, Paul Mclachlan, Fuyu Zou, Chenhua Shi, Lule Yu and Sheyda Kiani Mehr.

If youre interested in advancing this field with us, join our worldwide team of data scientists and AI specialists at GAIA.

In this Wired article, Turing Award winner Yoshua Bengio shares why deep learning must begin to understand the why before it can replicate true human intelligence.

In this technical overview of causal inference in statistics, find out whats needed to evolve AI from traditional statistical analysis to causal analysis of multivariate data.

This journal essay from 1999 offers an introduction to the Causal Markov Condition.

Read more:
Overview of causal inference in machine learning - Ericsson

To print this article, all you need is to be registered or login on Mondaq.com.

This article is the first of a five-part series of articlesdealing with what patentability of machine learning looks like in2019. This article begins the series by describing the USPTO's2019 Revised Patent Subject MatterEligibility Guidance (2019 PEG) in the context of the U.S.patent system. Then, this article and the four followingarticles will describe one of five cases in whichExaminer's rejections under Section 101 were reversed bythe PTAB under this new 2019 PEG. Each of the five cases discusseddeal with machine-learning patents, andmay provide some insight into how the 2019 PEG affects the patentability ofmachine-learning, as well as software more broadly.

The US patent laws are set out in Title 35 of the United StatesCode (35 U.S.C.). Section 101 of Title 35 focuses on severalthings, including whether the invention is classified aspatent-eligible subject matter. As a general rule, an invention isconsidered to be patent-eligible subject matter if it "fallswithin one of the four enumerated categories of patentable subjectmatter recited in 35 U.S.C. 101 (i.e.,process, machine, manufacture, or composition of matter)."1 This,on its own, is an easy hurdle to overcome. However, there areexceptions (judicial exceptions). These include (1) laws of nature;(2) natural phenomena; and (3) abstract ideas. If the subjectmatter of the claimed invention fits into any of these judicialexceptions, it is not patent-eligible, and a patent cannot beobtained. The machine-learning and software aspects of a claim face101 issues based on the "abstract idea" exception, andnot the other two.

Section 101 is applied by Examiners at the USPTO in determiningwhether patents should be issued; by district courts in determiningthe validity of existing patents; in the Patent Trial and Appeal Board(PTAB) in appeals from Examinerrejections, in post-grant-review (PGR)proceedings, and in covered-business-method-review(CBM) proceedings; and in the Federal Circuit on appeals. ThePTAB is part of the USPTO, and may hear an appeal of anExaminer's rejection of claims of a patent application when theclaims have been rejected at least twice.

In determining whether a claim fits into the "abstractidea" category at the USPTO, the Examiners and the PTAB mustapply the 2019 PEG, which is described in the following section ofthis paper. In determining whether a claim is patent-ineligible asan "abstract idea" in the district courts and the FederalCircuit, however, the courts apply the "Alice/Mayo" test;and not the 2019 PEG. The definition of "abstract idea"was formulated by the Alice and Mayo Supreme Court cases. Thesetwo cases have been interpreted by a number of Federal Circuitopinions, which has led to a complicated legal framework that theUSPTO and the district courts must follow.2

The USPTO, which governs the issuance of patents, decided thatit needed a more practical, predictable, and consistent method forits over 8,500 patent examiners to apply when determining whether aclaim is patent-ineligible as an abstract idea.3 Previously, theUSPTO synthesized and organized, for its examiners to compare to anapplicant's claims, the facts and holdings of each FederalCircuit case that deals with section 101. However, the large andstill-growing number of cases, and the confusion arising from"similar subject matter [being] described both as abstract andnot abstract in different cases,"4 led to issues.Accordingly, the USPTO issued its 2019 Revised Patent SubjectMatter Eligibility Guidance on January 7, 2019 (2019 PEG), whichshifted from the case-comparison structure to a new examinationstructure.5 The new examination structure,described below, is more patent-applicant friendly than the priorstructure,6 thereby having the potential toresult in a higher rate of patent issuances. The 2019 PEG does notalter the federal statutory law or case law that make up the U.S.patent system.

The 2019 PEG has a structure consisting of four parts: Step 1,Step 2A Prong 1, Step 2A Prong 2, and Step 2B. Step 1 refers to thestatutory categories of patent-eligible subject matter, while Step2 refers to the judicial exceptions. In Step 1, the Examiners mustdetermine whether the subject matter of the claim is a process,machine, manufacture, or composition of matter. If it is, theExaminer moves on to Step 2.

In Step 2A, Prong 1, the Examiners are to determine whether theclaim "recites" a judicial exception includinglaws of nature, natural phenomenon, and abstract ideas. Forabstract ideas, the Examiners must determine whether the claimfalls into at least one of three enumerated categories: (1)"mathematical concepts" (mathematical relationships,mathematical formulas or equations, mathematical calculations); (2)"certain methods of organizing human activity"(fundamental economic principles or practices, commercial or legalinteractions, managing personal behavior or relationships orinteractions between people); and (3) "mental processes"(concepts performed in the human mind: encompassing acts people canperform using their mind, or using pen and paper). These threeenumerated categories are not mere examples, but arefully-encompassing. The Examiners are directed that "[i]n therare circumstance in which they believe[] a claim limitation thatdoes not fall within the enumerated groupings of abstract ideasshould nonetheless be treated as reciting an abstract idea,"they are to follow a particular procedure involving providingjustifications and getting approval from the Technology CenterDirector.

Next, if the claim limitation "recites" one of theenumerated categories of abstract ideas under Prong 1 of Step 2A,the Examiner is instructed to proceed to Prong 2 of Step 2A. InStep 2A, Prong 2, the Examiners are to determine if the claim is"directed to" the recited abstract idea. In this step,the claim does not fall within the exception, despite reciting theexception, if the exception is integrated into a practicalapplication. The 2019 PEG provides a non-exhaustive list ofexamples for this, including, among others: (1) an improvement inthe functioning of a computer; (2) a particular treatment for adisease or medical condition; and (3) an application of "thejudicial exception in some other meaningful way beyond generallylinking the use of the judicial exception to a particulartechnological environment, such that the claim as a whole is morethan a drafting effort designed to monopolize theexception."

Finally, even if the claim recites a judicial exception underStep 2A Prong 1, and the claim is directed to the judicialexception under Step 2A Prong 2, it might still be patent-eligibleif it satisfies the requirement of Step 2B. In Step 2B, theExaminer must determine if there is an "inventiveconcept": that "the additional elements recited in theclaims provide[] 'significantly more' than the recitedjudicial exception." This step attempts to distinguish betweenwhether the elements combined to the judicial exception (1)"add[] a specific limitation or combination of limitationsthat are not well-understood, routine, conventional activity in thefield"; or alternatively (2) "simply append[]well-understood, routine, conventional activities previously knownto the industry, specified at a high level of generality."Furthermore, the 2019 PEG indicates that where "an additionalelement was insignificant extra-solution activity, [the Examiner]should reevaluate that conclusion in Step 2B. If such reevaluationindicates that the element is unconventional . . . this finding mayindicate that an inventive concept is present and that the claim isthus eligible."

In summary, the 2019 PEG provides an approach for the Examinersto apply, involving steps and prongs, to determine if a claim ispatent-ineligible based on being an abstract idea. Conceptually,the 2019-PEG method begins with categorizing the type of claiminvolved (process, machine, etc.); proceeds to determining if anexception applies (e.g., abstract idea); then, if an exceptionapplies, proceeds to determining if an exclusion applies (i.e.,practical application or inventive concept). Interestingly, thePTAB not only applies the 2019 PEG in appeals from Examinerrejections, but also applies the 2019 PEG in its other Section-101decisions, including CBM review and PGRs.7 However, the 2019PEG only applies to the Examiners and PTAB (the Examiners and thePTAB are both part of the USPTO), and does not apply to districtcourts or to the Federal Circuit.

Case 1: Appeal 2018-0074438 (Decided October 10,2019)

This case involves the PTAB reversing the Examiner's Section101 rejections of claims of the 14/815,940 patent application. Thispatent application relates to applying AI classificationtechnologies and combinational logic to predict whether machinesneed to be serviced, and whether there is likely to be equipmentfailure in a system. The Examiner contended that the claims fitinto the judicial exception of "abstract idea" because"monitoring the operation of machines is a fundamentaleconomic practice." The Examiner explained that "thelimitations in the claims that set forth the abstract idea are:'a method for reading data; assessing data; presenting data;classifying data; collecting data; and tallying data.'"The PTAB disagreed with the Examiner. The PTAB stated:

Specifically, we do not find 'monitoring the operation ofmachines,' as recited in the instant application, is afundamental economic principle (such as hedging, insurance, ormitigating risk). Rather, the claims recite monitoring operation ofmachines using neural networks, logic decision trees, confidenceassessments, fuzzy logic, smart agent profiling, and case-basedreasoning.

As explained in the previous section of this paper, the 2019 PEGset forth three possible categories of abstract ideas: mathematicalconcepts, certain methods of organizing human activity, and mentalprocesses. Here, the PTAB addressed the second of these categories.The PTAB found that the claims do not recite a fundamental economicprinciple (one method of organizing human activity) because theclaims recite AI components like "neural networks" in thecontext of monitoring machines. Clearly, economic principles and AIcomponents are not always mutually exclusive concepts.9 Forexample, there may be situations where these algorithms are applieddirectly to mitigating business risks. Accordingly, the PTAB waslikely focusing on the distinction between monitoring machines andmitigating risk; and not solely on the recitation of the AIcomponents. However, the recitation of the AI components did notseem to hurt.

Then, moving on to another category of abstract ideas, the PTABstated:

Claims 1 and 8 as recited are not practically performed in thehuman mind. As discussed above, the claims recite monitoringoperation of machines using neural networks, logic decision trees,confidence assessments, fuzzy logic, smart agent profiling, andcase-based reasoning. . . . [Also,] claim 8 recites 'an outputdevice that transforms the composite prediction output intohuman-readable form.'

. . . .

In other words, the 'classifying' steps of claims 1 and'modules' of claim 8 when read in light of theSpecification, recite a method and system difficult and challengingfor non-experts due to their computational complexity. As such, wefind that one of ordinary skill in the art would not find itpractical to perform the aforementioned 'classifying' stepsrecited in claim 1 and function of the 'modules' recited inclaim 8 mentally.

In the language above, the PTAB addressed the third category ofabstract ideas: mental processes. The PTAB provided that the claimdoes not recite a mental process because the AI algorithms, basedon the context in which they are applied, are computationallycomplex.

The PTAB also addressed the first of the three categories ofabstract ideas (mathematical concepts), and found that it does notapply because "the specific mathematical algorithm or formulais not explicitly recited in the claims." Requiring that amathematical concept be "explicitly recited" seems to bea narrow interpretation of the 2019 PEG. The 2019 PEG does notrequire that the recitation be explicit, and leaves the mathcategory open to relationships, equations, or calculations. Fromthis, the PTAB might have meant that the claims list a mathematicalconcept (the AI algorithm) by its name, as a component of theprocess, rather than trying to claim the steps of the algorithmitself. Clearly, the names of the algorithms are "explicitlyrecited"; the steps of the AI algorithms, however, are notrecited in the claims.

Notably, reciting only the name of an algorithm, rather thanreciting the steps of the algorithm, seems to indicate that theclaims are not directed to the algorithms (i.e., the claims have apractical application for the algorithms). It indicates that theclaims include an algorithm, but that there is more going on in theclaim than just the algorithm. However, instead of determining thatthere is a practical application of the algorithms, or an inventiveconcept, the PTAB determined that the claim does not even recitethe mathematical concepts.

Additionally, the PTAB found that even if the claims hadbeen classified as reciting an abstract idea, as the Examiner hadcontended the claims are not directed to that abstractidea, but are integrated into a practical application. The PTABstated:

"Appellant's claims address a problem specificallyusing several artificial intelligence classification technologiesto monitor the operation of machines and to predict preventativemaintenance needs and equipment failure."

The PTAB seems to say that because the claims solve a problemusing the abstract idea, they are integrated into a practicalapplication. The PTAB did not specify why the additional elementsare sufficient to integrate the invention. The opinion actuallydoes not even specifically mention that there are additionalelements. Instead, the PTAB's conclusion might have been that,based on a totality of the circumstances, it believed that theclaims are not directed to the algorithms, but actually just applythe algorithms in a meaningful way. The PTAB could have fit thisreasoning into the 2019 PEG structure through one of the Step 2A,Prong 2 examples (e.g., that the claim applies additional elements"in some other meaningful way"), but did not expressly doso.

This case illustrates:

(1) the monitoring of machines was held to not be an abstractidea, in this context;(2) the recitation of AI components such as "neuralnetworks" in the claims did not seem to hurt for arguing anyof the three categories of abstract ideas;(3) complexity of algorithms implemented can help with the"mental processes" category of abstract ideas; and(4) the PTAB might not always explicitly state how the rule for"practical application" applies, but seems to apply itconsistently with the examples from the 2019 PEG.

The next four articles will build on this background, and willprovide different examples of how the PTAB approaches reversingExaminer 101-rejections of machine-learning patents under the 2019PEG. Stay tuned for the analysis and lessons of the next case,which includes methods for overcoming rejections based on the"mental processes" category of abstract ideas, on anapplication for a "probabilistic programming compiler"that performs the seemingly 101-vulnerable function of"generat[ing] data-parallel inference code."

Footnotes

1 MPEP 2106.04.

2Accordingly, the USPTO must follow both the Federal Circuit'scase law that interprets Title 35 of the United States Code, andmust follow the 2019 PEG. The 2019 PEG is not the same as theFederal Circuit's standard the 2019 PEG does notinvolve distinguishing case law (the USPTO, in its 2019 PEG, hasdeclared the Federal Circuit's case law to be too clouded to bepractically applied by the Examiners. 84 Fed. Reg. 52.). The USPTOpractically could not, and actually did not, synthesize theholdings of each of the Federal Circuit opinions regarding Section101 into the standard of the 2019 PEG. Therefore, logically, theonly way to ensure that the 2019 PEG does not impinge on thestatutory rights (provided by 35 U.S.C.) of patent applicants, asinterpreted by the Federal Circuit, is for the 2019 PEG to definethe scope of the 101 judicial exceptions more narrowly than theStatutory requirement. However, assuming there are instances wherethe 2019 PEG defines the 101 judicial exceptions more broadly thanthe statutory standard (if the USPTO rejects claims that theFederal Circuit would not have), that patent applicant may haveadditional arguments for eligibility.

3 84 Fed.Reg. 50, 52.

4Id.

5 TheUSPTO also, on October 17 of 2019, issued an update to the 2019PEG. The October update is consistent with the 2019 PEG, and merelyprovides clarification to some of the terms used in the 2019 PEG,and clarification as to the scope of the 2019 PEG. October 2019 Update: Subject MatterEligibility (October 17, 2019), https://www.uspto.gov/sites/default/files/documents/peg_oct_2019_update.pdf.

6See "Frequently Asked Questions (FAQs) on the 2019Revised Patent Subject Matter Eligibility Guidance ('2019PEG')", C-6 (https://www.uspto.gov/sites/default/files/documents/faqs_on_2019peg_20190107.pdf)("Any claim considered patent eligible under the currentversion of the MPEP and subsequent guidance should be consideredpatent eligible under the 2019 PEG. Because the claim at issue wasconsidered eligible under the current version of the MPEP, theExaminer should not make a rejection under 101 in view ofthe 2019 PEG.").

7See American Express v. Signature Systems, CBM2018-00035(Oct. 30, 2019); Supercell Oy v. Gree, Inc., PGR2018-00061 (Oct.15, 2019).

8 https://e-foia.uspto.gov/Foia/RetrievePdf?system=BPAI&flNm=fd2018007443-10-10-2019-0.

9Notably, the "mental process" category and notthe "certain methods of organizing human activity"category is the one that focuses on the complexity of theprocess. Furthermore, as shown in the following paragraph, the"mental process" category was separately discussed by thePTAB, again mentioning the algorithms. Accordingly, the PTAB islikely not mentioning the algorithms for the purpose of describingthe complexity of the method.

The content of this article is intended to provide a generalguide to the subject matter. Specialist advice should be soughtabout your specific circumstances.

View original post here:
Machine Learning Patentability In 2019: 5 Cases Analyzed And Lessons Learned Part 1 - Mondaq News Alerts

When it comes to identifying early cyber threats, its important to have laser-like precision. Mapping out a threat environment can be done with a range of approaches, and a team of researchers from Purdue University created a new system for just such applications. They are calling that approach LIDAR, or lifelong, intelligent, diverse, agile and robust.

This is not to be confused with LiDAR, for Light Detection and Ranging, a kind of remote sensing system that uses laser pulses to measure distances from the sensor. The light-specific LiDAR, sometimes also written LIDAR, is a valuable tool for remote sensing and mapping, and features prominently in the awareness tools of self-driving vehicles.

Purdues LIDAR, instead, is a kind of architecture for network security. It can adapt to threats, thanks in part to its ability to learn three ways. These include supervised machine learning, where an algorithm looks at unusual features in the system and compares them to known attacks. An unsupervised machine learning component looks through the whole system for anything unusual, not just unusual features that resemble attacks. These two machine-learning components are mediated by a rules-based supervisor.

One of the fascinating things about LIDAR is that the rule-based learning component really serves as the brain for the operation, said Aly El Gamal, an assistant professor of electrical and computer engineering in Purdues College of Engineering. That component takes the information from the other two parts and decides the validity of a potential attack and necessary steps to move forward.

By knowing existing attacks, matching to detected threats, and learning from experience, this LIDAR system can potentially offer a long-term solution based on how the machines themselves become more capable over time.

Aiding the security approach, said the researchers, is the use of a novel curiosity-driven honeypot, which can like a carnivorous pitcher plant lure attackers and then trap them where they will do no harm. Once attackers are trapped, it is possible the learning algorithm can incorporate new information about the threat, and adapt to prevent future attacks making it through.

The research team behind this LIDAR approach is looking to patent the technology for commercialization. In the process, they may also want to settle on a less-confusing moniker. Otherwise, we may stumble into a future where users securing a network of LiDAR sensors with LIDAR have to enact an entire Whos on First? routine every time they update their cybersecurity.

See more here:
New cybersecurity system protects networks with LIDAR, no not that LiDAR - C4ISRNet