Equality of opportunity in supervised learning

Equality of opportunity in supervised learning Hardt et al., NIPS’16

With thanks to Rob Harrop for highlighting this paper to me.

There is a a lot of concern about discrimination and bias entering our machine learning models. Today’s paper choice introduces two notions of fairness: equalised odds, and equalised opportunity, and shows how to construct predictors that are fair under these criteria. One very appealing feature of the model is that in the case of uncertainty caused by under-representation in the training data, the cost of less accurate decision making in that demographic is moved from the protected class (who might otherwise for example not be offered loans), to the decision maker. I’m going to approach the paper backwards, and start with the case study, as I find a motivating example really helps with the intuition.

Loans, race, and FICO scores

We examine various fairness measures in the context of FICO scores with the protected attribute of race. FICO scores are a proprietary classifier widely used in the United States to predict credit worthiness. Our FICO data is based on a sample of 301,536 TransUnion TransRisk scores from 2003.

We’re interesting in comparing scores, the Continue reading

Amateur Radio and FT8

My interest in SDR got me into Amateur Radio. One reason was that so that I could transmit on non-ISM bands and with more power. Turns out the 2.3GHz band available to Amateur Radio operators is much quieter than the 2.4GHz band where WiFi, bluetooth, microwave ovens, drones, cordless phones and everything else lives. Shocker, I know.

Amateur radio doesn’t just have voice and morse code, there’s also digital modes.

A popular mode is FT8. It’s only used to exchange signal reports, so there’s no chatting. It’s in fact often practically unattended.

It’s a good way to check the quality of your radio setup, and the radio propagation properties that depend on how grumpy the ionosphere is at the moment.

If you transmit, even if you nobody replies, you’ll be able to see on PSKReporter who heard you, which is useful.

Because propagation should be pretty much symmetric, receiving a strong signal should mean that two-way communication is possible with the station. Though FT8 is a slow mode that will get through where others won’t, so a weak FT8 signal means that any voice communication is very unlikely to get through.

Unfortunately unlike WSPR the standard FT8 Continue reading

Active Directory & Ansible Tower

Ansible-Get-Started-Windows

 Welcome to the second installment of our Windows-centric Getting Started series!

Last time we walked you through how Ansible connects to a Windows host. We’ve also previously explored logging into Ansible Tower while authenticating against an LDAP directory. In this post, we’ll go over a few ways you can use Ansible to manage Microsoft’s Active Directory. Since AD plays a role in many Windows environments, using Ansible to manage Windows will probably include running commands against the Active Directory domain.


First, Set Your Protocol

We’ll be using WinRM to connect to Windows hosts, so this means making sure Ansible or Tower knows that. Machine credentials in Ansible Tower can be created and used along with variables, but when using Ansible in a terminal the playbook should make it clear with variables:

---
- name: Your Windows Playbook
  hosts: win
  vars:
    ansible_ssh_user: administrator
    ansible_ssh_pass: ThisIsWhereStrongPassesGo
    ansible_connection: winrm
    ansible_winrm_server_cert_validation: ignore

- tasks:

Along with using the local admin account/pass, the WinRM connection method is named specifically. The variable to ignore the certificate validation is for standalone, non-domain hosts because a domain-joined instance should have certificates validated on the domain.


Where’s the Domain?

Speaking of domains, Ansible can spin up a new domain Continue reading

OpenDKIM-OpenDMARC and a Chrooted Postfix Instance

No Postfix installation is complete without OpenDKIM and OpenDMARC.

While some people go for all-in-one solutions that does all of these for them with a single command or two (and then cry to their gods as soon as the system fails as they have no idea how to debug it), the rest of us rather to be our own boss and set things up manually and carefully based on our needs, so we can troubleshoot it if when things go wrong.

This however, is easier to be said than done. In this post, rather than trying to explain what they are and how they can be set up (which can be found everywhere on the web), I am mainly going to address the issues that you might encounter when running your Postfix and these Milters on the same system running Ubuntu.

Background

OpenDKIM and OpenDMARC are designed to be used as Milters. They are two different programs for two different -and yet related- tasks.

They show a lot of similarities in their configuration files and both suffer from the same limitations when running along with a chrooted Postfix instance.

While in a recent enough version of Postfix, daemons are Continue reading

Learning TrustSec – An Introduction to Inline Tagging

In my last article, Basic TrustSec – Implementing Manual SGTs and SGACLs,
 we talked about a basic TrustSec configuration. In that example, we shared the understanding of having two devices connected to a single switch and enforcing traffic policies via SGACL. We know that there are more scalable and automated ways to configure TrustSec enabled networks, but our goal is to work toward understanding the building blocks.

In today’s article, we will expand our knowledge and connect the two devices to different switches. The trunks between these switches will be configured to carry the associated source SGT’s (Security Group Tags). The topology used for this discussion is as follows.

Topology

To demonstrate the topic of inline SGT, we will need to accomplish the following.

  1. Configure and Confirm that 192.168.254.11 (connected to c9kSW1) is recognized by its switch with an SGT of 2.
  2. Configure and Confirm that 192.168.254.100 (connected to c9kSW2) is recognized by its switch with an SGT of 3.
  3. Configure the trunk between the switches to carry SGTs
  4. Configure an enforcement policy to demonstrate overall functionality

Configuration Steps

c9kSW1 configuration/confirmation for host port

//We are using static SGT and need to do IP Device  Continue reading

Learning TrustSec – An Introduction to Inline Tagging

In my last article, Basic TrustSec – Implementing Manual SGTs and SGACLs,
 we talked about a basic TrustSec configuration. In that example, we shared the understanding of having two devices connected to a single switch and enforcing traffic policies via SGACL. We know that there are more scalable and automated ways to configure TrustSec enabled networks, but our goal is to work toward understanding the building blocks.

In today’s article, we will expand our knowledge and connect the two devices to different switches. The trunks between these switches will be configured to carry the associated source SGT’s (Security Group Tags). The topology used for this discussion is as follows.

Topology

To demonstrate the topic of inline SGT, we will need to accomplish the following.

  1. Configure and Confirm that 192.168.254.11 (connected to c9kSW1) is recognized by its switch with an SGT of 2.
  2. Configure and Confirm that 192.168.254.100 (connected to c9kSW2) is recognized by its switch with an SGT of 3.
  3. Configure the trunk between the switches to carry SGTs
  4. Configure an enforcement policy to demonstrate overall functionality

Configuration Steps

c9kSW1 configuration/confirmation for host port

//We are using static SGT and need to do IP Device  Continue reading

Learning TrustSec – An Introduction to Inline Tagging

In my last article, Basic TrustSec – Implementing Manual SGTs and SGACLs,
 we talked about a basic TrustSec configuration. In that example, we shared the understanding of having two devices connected to a single switch and enforcing traffic policies via SGACL. We know that there are more scalable and automated ways to configure TrustSec enabled networks, but our goal is to work toward understanding the building blocks.

In today’s article, we will expand our knowledge and connect the two devices to different switches. The trunks between these switches will be configured to carry the associated source SGT’s (Security Group Tags). The topology used for this discussion is as follows.

Topology

To demonstrate the topic of inline SGT, we will need to accomplish the following.

  1. Configure and Confirm that 192.168.254.11 (connected to c9kSW1) is recognized by its switch with an SGT of 2.
  2. Configure and Confirm that 192.168.254.100 (connected to c9kSW2) is recognized by its switch with an SGT of 3.
  3. Configure the trunk between the switches to carry SGTs
  4. Configure an enforcement policy to demonstrate overall functionality

Configuration Steps

c9kSW1 configuration/confirmation for host port

//We are using static SGT and need to do IP Device  Continue reading

MTU (Maximum Transmit Unit) and MSS (Maximum Segment Size)

What is difference  between MTU and MSS ? Most frequent question asked on the internet by networking guys.Hope this post will answer all queries related to MTU and MSS. As per Wikipedia , the maximum transmission unit (MTU) is the size of the largest protocol data unit (PDU) that can be communicated in a single, network layer, transaction , that […]

Zero Trust. Maximize Network Virtualization and Micro-segmentation

 

It’s official: when it comes to security threats, the question IT teams should be asking is not if but when. VMware recently commissioned Forrester Consulting to evaluate how organizations are improving the security of their infrastructure through network virtualization and micro-segmentation. Analysis found that 92% of respondents reported having faced minor security incidents in the last 12 months alone, while 65% of respondents endured a major incident in the same time span. These figures seal the deal; the naïve days of preparing for potential issues are long gone. Cyber threats are real, imminent, and happen often.

 

 

Companies today attribute more of their security issues to improper network segmentation than to the volume of threats overall. In response, leaders across industries are turning to network virtualization – specifically the Zero Trust security model – as a key strategy in combating threats. This strategy posits that whether a network is labeled secure or insecure, both should be treated as equally vulnerable. Further, the Zero Trust model supports the argument that ”traditional, perimeter-based security configurations are no longer a sufficient measure for protecting the network, and highlights steps companies can take to better secure their network, starting with network virtualization Continue reading

Link Propagation 117

Welcome to Link Propagation, a Packet Pushers newsletter. Link Propagation is included in your free membership. Each week we scour the InterWebs to find the most relevant practitioner blog posts, tech news, and product announcements. We drink from the fire hose so you can sip from a coffee cup. Blogs How Real Life Can Change […]

Transitioning Away From Legacy IT

One of the more exciting things I saw at Dell Technologies World this week was the announcement by VMware that they are supporting Microsoft Azure now in additional to AWS. It’s interesting because VMware is trying to provide a proven, stable migration path for companies that are wanting to move to the cloud but still retain their investments in VMware and legacy virtualization. But is offing legacy transition a good idea?

Hold On For One More Day

If I were to mention VLAN 1002-1005 to networking people, they would likely jump up and tell me that I was crazy. Because those VLANs are not valid on any Cisco switches save for the Nexus line. But why? What makes these forbidden? Unless you’re studying for your CCIE you probably just know these are bad and move on.

Turns out, they are a legacy transition mechanism from the IOS-SX days. 1002 and 1004 were designed to bridge FDDI-to-Ethernet, and 1003 and 1005 did the same for Token Ring. As Greg Ferro points out here, this code was tightly bound into IOS-SX and likely couldn’t be removed for fear of breaking the OS. The reservation continued forward in all IOS branches except Continue reading

1 … 1,573 1,574 1,575 1,576 1,577 … 3,795