Following are a few presentations. You can download them from the “Latest Thinking” section of my blog: https://asiliconvalleyinsider.com/latest-thinking/ or from the links below.
While we hear a lot about generating text and generating images, we do not as often hear applications and technologies about generating voice. Nevertheless, from time to time, we hear some new models for generating music. So is there a market for generative voice and in particular voice cloning? What could be the modeling technologies and features of a voice cloning product? What could be the go-to market for such a product? When there is a lot of hype and confusion in a market, such as is the case today for generative AI, developing and launching a new product in that market is not so easy. Having said that here are a few answers to those three questions:
Read More: Generative Voice and Voice Cloning
Machine learning models have many applications in finance to both improve the bottom and top line of an enterprise. Models can streamline workflows, provide insights, help with decision-making, and increase customer engagement. Multiple models can be implemented together for an end-to-end workflow such as a complete solution to automate accounts payable from data extraction, accounts coding, anomaly detection, to bank reconciliation. More complex models can also be investigated such as recommender systems to provide personalized experiences, reinforcement learning for learning user feedback, and graph neural networks for demand forecasting.
Read more: Machine Learning for Finance
While A/B testing aims to establish any difference between two randomized groups “A” and “B”, and the causal effect of changing “x” on the outcome “y” on group “A” versus group “B”, causal inference techniques attempt to discover and mitigate the cofounding factors “c” (called also omitted variable bias in statistics) that affect the input of interests “x” with the outcome of interest “y”. Knowing the relationships between “x” and “y” enables generating actionable insights in order to make the right course of action to move from "x" to "y".
Read more: Causal Inference
One of the key goals of many data scientist teams is to develop models faster for new use cases and new applications. To that end, “transfer learning” which aims to train on one task, and to transfer that learning to a new task has been widely used in particular, in computer vision, and language models. However, two new techniques have emerged recently: “multi-task learning” which aims to train a system on many tasks, and transfer that learning to the system for a new task, and “meta-learning” which empowers a system to learn from many tasks, and transfer that learning to the system for similar tasks.
Read more: Multi-Task & Meta Learning
The Internet of Things (IoT) is all about machine learning at scale. The value of an IoT solution is in its machine learning applications. Those applications are what will motivate the IoT users to be more engaged with their devices. And, the more devices are connected to the IoT platform, the more the machine learning applications will provide value to the users of the IoT devices.
Read more: Machine Learning for IoT
Every mobile platform has its own development environment: BlackBerry (Java and C/C++), iOS (Objective C), and Android (Dalvik Java). Fortunately for developers, there are a few alternatives. Some tools can wrap existing HTML into native libraries such as PhoneGap or even generate native code from existing HTML such as Appcelerator.
Read more: Cross Mobile Development
IDS is dead. Welcome IDP. IPSec is dead. Welcome SSL. Well, history repeats itself. IDS and IDP are two different technologies for two different applications. The same is true for IPSec and SSL. IPSec provides data privacy at the network layer. SSL provides data privacy at the application layer. There will always be the need for an enterprise network to have IPSec and SSL to co-exist like there is for IDS and IDP.
Read more: IPSec
MPLS emerged as a traffic engineering technology to optimize traffic loads on the Internet backbone by distributing loads evenly in the backbone. An ingress system shares its transit load across multiple paths. With more and more functions, features, and applications being developed by the IETF and the MPLS forum, MPLS is enabling a new range of applications beyond traffic engineering: Differentiated Services, Layer 2 VPNs/VPLS, Layer 3 VPNs/BGP MPLS, Multi-Service transport for Ethernet/Frame/ATM.
Read more: Migrating to MPLS
G-MPLS initial goal was to facilitate the required interaction between the IP and optical domains by extending MPLS traffic engineering for optical networks. Now, G-MPLS aims to create a universal control plane for dynamic provisioning and restoration of optical transport networks and will be key for the architecture of future core optical networks evolving towards meshes of DWDM transport systems connecting Optical and Photonic Switches (OXC/PXC).
Read more: GMPLS
The communication industry is moving to a horizontal integration like the computing industry did in the early 90s. The impact of the 1996 Telecom Act and the Internet are causing the destruction of the overall market value of the industry. Traditional Telecom Services are competing with the Internet, and Wireline Services are competing with Wireless Services.
Read more: The Communication Industry
Although corporations agree that Internet security should be one of the enabling technologies for the business use of the Net, the market has still not taken off. Why? What are the market dynamics in Intrusion Detection, the latest Internet security market segment after the explosion of the firewall market?
Read more: The Internet Security IDS Market