Frequently Asked Questions
What is Intraspexion?
Intraspexion is a private company organized as a Delaware for-profit corporation, with headquarters in the state of Washington. Intraspexion offers a patented software system to identify litigation (and other) risks. Intraspexion owns seven (7) patents for its innovations in the context of words that are used in specific classifications and “deep learning” neural networks.
What is “Deep Learning”?
Deep Learning is the informal name for a multi-layer neural network. As the result of academic breakthroughs beginning in 2013, such neural networks have enabled computers “to learn” from either images or words or both. In the case of words and “supervised learning,” a deep learning neural network is now able to learn the patterns of words (and keep them in context) when there are a sufficient number of examples of the category to be learned.
How inventive have you been? In other words, how does the number of your issued patents stack up against others?
Currently, as of July 20, 2018, we're in 5th place, and we're the only startup in the Top 10. For this result, we searched the USPTO for patents with "deep learning" or "deep neural" in Claim(s) ranging from January 1, 2013 through July 10, 2018.
Top 10 Results
1. Google - 24
2. IBM - 18
3. Microsoft - 14
4. Siemens Healthcare GmbH - 10
5. Intraspexion - 7
6. Adobe Systems Incorporated - 5
6. Amazon.com, Amazon Technologies, and subsidiary A9.com - 5
6. NEC - 5
9. Mitsubishi - 4
9. Electronics and Technology Research Institute (KR) - 4
How Accurate Is Your System?
The only accuracy we can report pertains to the one-and-only publicly available dataset, Enron. For our first category of litigation risk, we chose "employment discrimination." Enron was known for fraud, not discrimination, but we found some True Positive "needles" in the Enron haystack.
Then we added 10,000 Enron non-discrimination emails to the unrelated set, so the model could “understand” English in the context of emails.
Then, as a proxy for the internal company emails "from yesterday," we looked at a held-out set of 20,401 Enron emails that our system had not previously assessed.
Result: Our Employment Discrimination "model" called out 25 emails as being "related" to discrimination. Our previously-found four (4) True Positives, i.e., the "needles," were among the 25.
That's 25 out of 20,401 emails, a fraction of 0.001225; and that's a little less than one-eighth of one percent. Our System said “pass” as to about 99.88% of the emails it processed and "take a look" at 0.12%.
Besides "employment discrimination," What Other Litigation Threats (Types of Cases) Does Your System Detect?
Currently, our system provides insight to "employment discrimination" risks. As litigators will appreciate, there are many such business-relevant categories in PACER. And, for each category, there are usually a large number of examples; i.e., lawsuits filed in that category. We can train a Deep Learning model for any business-relevant litigation category.
What About the Privacy Concerns of Employees?
Where a “computer technology resource" (CTR) policy is in place, and the employee is aware of the CTR policy, and agrees to it, our general understanding is that there is no reasonable expectation of privacy when an employee is using a company-owned computer. See Holmes v. Petrovich Development, LLC, 191 CAl.App.4th 1047, 119 Cal.Rptr.3d 878 (2011) (holding that not even the attorney-client privilege can be raised against a company when a CTP policy is in place and the company monitors the emails flowing through its own computer resources).
Does Intraspexion store or forward any company emails?
Do You Have Office 365 Connectivity?
Do You Have a Cloud-Based Solution?
We are exploring "cloud" and have more than one candidate in mind. However, based on initial customer interest, our first deployments are for on-premises solutions.
How Secure Is Your System?
For Intraspexion’s on-premises solution, your data never goes beyond our customer's firewall. Our System is as secure as that. When we have a cloud solution as well, the System security may be as good or even better.
How Does the Free Trial Work?
STEP 1: We sign your NDA and a Trial Agreement.
STEP 2: You find and reserve a war room for the trial.
STEP 3: We'll send you a laptop and a server. The server will have a GPU card in it that's pre-trained for "discrimination," along with set-up instructions for you to use "on premises" for a limited time. You pay only for shipping.
STEP 4: You'll set the hardware up in the war room, and maybe ask IT for help, but set the hardware up in a way that (for starters) is not connected to your internal email system. This way, you'll isolate our laptop-server combination away from your company's email system.
Then you can try our system in two ways: (1) By putting in some Enron emails, which are available from the EDRM.net; and (2) By putting in emails (in .pst format) from the production set(s) from one or more of your now-closed discrimination cases where you have already identified which of the emails were risky or were "smoking guns."
We'll both know what to do after that.
What Technical Assistance Do You Provide?
For installations, we provide step-by-step instructions. The instructions include both text and images.
What's Your Pricing Structure?
Each prospective customer has a unique litigation profile. Their prevention priorities differ. As a result, we have no pricing structure. We customize our pricing and we keep our relationship with each customer confidential.
Is There a Difference Between Intraspexion's Deep Learning and eDiscovery Software Tools?
Yes. Deep Learning is an AI technology that is unlike any eDiscovery software.
We conducted two searches to demonstrate that Deep Learning is not currently a tool in the eDiscovery toolbox.
The first search was of an eDiscovery standard source, the Electronic Discovery Reference Model, the EDRM (www.edrm.net). Using the local search engine, we searched for “deep learning” and found that phrase only in the Glossary.
We searched for “deep neural” as a shorthand for “deep neural network,” and found nothing.
We also searched for “neural network” and also found it in the Glossary, and in a 2010 article, “Develop Review Strategy/Plan,” in Section 1.10.4. There was one mention of “neural network” and it was contained within this single sentence: “Concept search technology may be based on neural networks, Baysian methods, latent semantic indexing, or other high-level mathematical algorithms designed to learn the underlying association of words.”
However, in each instance, there was no mention of any such product or tool.
For our second search, we went to Relativity.com, a platform which is familiar to just about every eDiscovery practitioner. We searched for “deep learning,” “deep neural,” “neural network,” and the plural, “neural networks.”
In each case, the Relativity local search engine produced the same result: “0 possible matches.”
Perhaps the key difference is this: eDiscovery technologies are used in the context of existing lawsuits and datasets obtained from custodians. Intraspexion’s Deep Learning system is an early warning system. It surfaces the risky emails that one might find in a lawsuit, but finds them in yesterday's emails and does so likely before a lawsuit is ever filed.
No eDiscovery tool has been designed or used for that purpose.
For the Sake of Customer Confidence and Transparency, Can You Tell Us Anything About the Software You Use?
Yes. While some of the software in our patented system is proprietary and we treat that coding as a trade secret, and our confidential intellectual property, we are proud to say that portions of our system are powered by agreement with dtSearch®. However, we don't use dtSearch for "search." We use it for indexing and for the visualization of words that experts have curated as being relevant to the subject matter. In addition, we use Google's open-sourced version of TensorFlow. TensorFlow, the TensorFlow logo and any related marks are trademarks of Google Inc. TensorFlow is included in our system under an Apache 2.0 license, the terms of which are incorporated herein in full by this reference.