July 26, 2020

Researchers develop new learning algorithm to boost AI efficiency

A working group led by two computer scientists Wolfgang Maass and Robert Legenstein of TU Graz has adopted this principle in the development of the new machine learning algorithm e-prop (short for e-propagation). Researchers at the Institute of Theoretical Computer Science, which is also part of the European lighthouse project Human Brain Project, use spikes in their model for communication between neurons in an artificial neural network. The spikes only become active when they are needed for information processing in the network. Learning is a particular challenge for such less active networks, since it takes longer observations to determine which neuron connections improve network performance. Previous methods achieved too little learning success or required enormous storage space. E-prop now solves this problem by means of a decentralized method copied from the brain, in which each neuron documents when its connections were used in a so-called e-trace (eligibility trace). The method is roughly as powerful as the best and most elaborate other known learning methods.

Data Leadership Book Review and Interview

The Data Leadership Framework is first about acknowledging that there is a whole bunch of stuff an organization needs to do to make the most of data. The five DLF Categories are where we evaluate an organization’s data capabilities and figure out where they are struggling most among the complexity. The twenty-five DLF Disciplines are where we then focus energy (i.e., invest our limited resources) to make the biggest outcomes. By creating relative balance across the DLF Categories, we maximize the overall impact of our data efforts. This is what we need to be doing all the time with data, but without something like the Data Leadership Framework, the problems can feel overwhelming and people have trouble figuring out where to start, or what to do next. This is true of everybody, from data architects and developers to the CEO. If we can use the Data Leadership Framework to make sense amidst the chaos, the individual steps themselves are much less daunting. Data competency is no longer a “nice-to-have” item. From data breaches to analytics-driven disruptors in every industry, this is as big of a deal to businesses as cash flow.

Enterprise Architecture for Managing Information Technology Standards

While globalization is excellent for business as it extends opportunities for markets that were previously closed and permits the sharing of ideas and information across different platforms, it could threaten the budgetary plans of SMBs. Investments in licensing, infrastructure, and global solutions, in general, hit this segment harshly. Lack of Talent Pool: This problem is primarily limited to the technology segment. Around half of employees lack the critical thinking skills that would qualify them to grow further in this field. The IT team faced the most significant hurdle so far is having members that aren’t smart enough to put a general hardware and software security environment cost-effectively. IT Policy Compliance Failure: Specific technologies used by IT projects don’t comply with the policy rules as defined by their departments. IT departments are sometimes unaware of techniques used by their teams and business stakeholders, increasing the risk of uncontrolled data flows and non-compliance. Besides, these technologies are sometimes incompatible with the existing portfolio. This increases IT debt, primarily if technology standards are not enforced.

IoT Architecture: Topology and Edge Compute Considerations

Network engineers often have experience with a particular topology and may assume it can be used in any setting, but sometimes another choice would be more optimal for a different use case. To determine whether a mesh networking topology is a good choice for your application, it is important to understand the pros and cons of this strategy. A critical factor to analyze is your system's timing requirements. Mesh networking topologies route data from node to node across a network that is architected in a mesh. So the "hops" need to be accounted for due to added latency. Do you need the data back in 100 mS or can you live with once a second? ... Wireless point-to-point (PTP) and point-to-multipoint (PTMP) are topologies used for connectivity in a wide range of applications, such as use cases where you want to replace cables with wireless communication. These protocols communicate between two devices (point-to-point) or from one device to many (point-to-multipoint). There are a few factors to consider, such as distance, timing and battery power that may indicate if a PTP network is needed versus a mesh network.

An introduction to confidential edge computing for IoT security

Recent attacks, even outside of IoT, showed that hackers exploited weak configurations of public cloud services to access sensitive data. The reason that hackers succeeded in obtaining sensitive information stored on a public cloud had nothing to do with the security mechanisms implemented by the cloud provider but were rather the result of little mistakes made by the end users, typically in the Web Application Firewall (WAF) that controls the access to the cloud network or by leaving credentials unprotected. These little mistakes are almost inevitable for companies that have a cloud-only infrastructure. However, by demarcating sensitive and non-sensitive information, this could help their IT teams in setting up the cloud services to achieve safer security practices. Those mistakes emphasize the need for a broader security expertise aiming at defining the security architecture to be enforced on the overall system and at finding out whether the security features of the cloud provider need to be completed by additional protection mechanisms. A first logical step consists of demarcating sensitive and non-sensitive information, to help the IT team establish appropriate priorities.

Read more here ...