Alibaba DAMO Academy’s predictions: prime 10 know-how traits in 2019 | Enterprise-IT alignment

0
0
Alibaba DAMO Academy's predictions: top 10 technology trends in 2019 | Business-IT alignment


From speech AI, super-large graph neural networks, heterogenous computing structure, to autonomous driving, blockchain and data-protection applied sciences, machine intelligence has been producing nice impacts on our lives, and we are able to count on an accelerated tempo of know-how revolution within the yr forward, in response to the Alibaba DAMO Academy:

Development 1: Actual-time city simulation turns into attainable

Extra sources will likely be allotted to applied sciences powering an clever “metropolis mind” and its purposes, whereas a metropolis simulation mannequin reflecting the real-time impulses and actions of a bodily metropolis could be constructed to facilitate the optimization of metropolis governance. Extra cities in China are anticipated to have a “metropolis mind” in 2019.

Development 2: Speech AI in sure areas to go Turing Take a look at

As speech intelligence know-how advances, real-time text-to-speech on cell gadgets could be virtually similar to human speech, even passing the Turing take a look at in sure conversations, akin to ones utilizing a robotic voice to alert about supply standing. Guidelines and rules must be established to manage the trade, to keep away from misusing the know-how in commercials, harassing cellphone calls and fraud.

Development 3: Specialised AI chips on the rise

Reminiscence-centric AI chip structure based mostly on the latest-generation of 3D-stacked reminiscence know-how will grow to be mainstream. As well as, brain-inspired neuromorphic computing chips will acquire momentum, whereas researchers search killer purposes. Specialised AI chips could problem the dominant place of GPUs for AI coaching in knowledge facilities. 

Development 4: Tremendous-large graph neural networks (GNNs) will equip machines with widespread sense

GNNs that mix with deep studying can hyperlink end-to-end studying and inductive reasoning collectively, which could have the ability to deal with points, akin to relational reasoning and explainability that couldn’t be tackled by deep studying earlier than. Strong GNNs can perform as a human mind that consists of neurons, equipping machines with widespread sense, comprehension and cognitive capabilities. 

Development 5: Restructuring of computing structure 

The heterogenous computing structure with FPGA, GPU and ASIC, together with rising reminiscence gadgets, are driving the evolution of conventional computing structure. Within the foreseeable future, the elemental modifications to computing structure, underscored by the transition from CPU-centered common computing to Area-specific computing pushed by purposes and know-how, will speed up the arrival of AI and the quantum-computing period.



Supply hyperlink

This site uses Akismet to reduce spam. Learn how your comment data is processed.