Medium Voltage Power Cables includes Single Core Medium Voltage Cable, Three Cores Medium Voltage Cable. They are used for power networks, industrial plants.
Generally, Medium Voltage Cables are Copper Wire conductor, Semi-conductor Shield, XLPE insulation, Semi-conductor Shield, Steel Wire Armored or Steel Tape Armored, with PVC sheath or LSOH Sheath.
We supply lots of those Cable as per IEC 60502 and other requirements. BDK is top supplier for China Southern Power Grid.
Medium Voltage Power Cables,Medium Voltage Cable,Cable Medium Voltage,Mv Cable Shenzhen Bendakang Cables Holding Co., Ltd , https://www.bdkcables.com
Wang Xiaogang, the Dean of Shangtang Research Institute, highlighted that in 2014, artificial intelligence (AI) face recognition technology achieved a breakthrough by surpassing human visual accuracy for the first time. However, even with this advancement, the error rate remains as high as one in a billion. He compared machine-based face recognition to a 4-bit password, suggesting that it has now evolved to an 8-bit level. This means the accuracy of facial recognition has improved by four orders of magnitude in just four years.
During a speech titled “The Present and Future of AI Empowerment†at the MIT Technology Review Summit, Wang discussed how AI is transforming everyday life through cloud computing and edge devices. He emphasized that deep learning has been the key driver behind AI's rapid development over the past decade. Three main factors have fueled this progress: big data, powerful cloud computing resources—including GPUs and AI chips—and advancements in deep learning algorithms.
For example, in 2014, machine face recognition had an error rate of 0.01%, but today it can be as low as 0.0000001%. This dramatic improvement reflects the evolution from a basic 4-bit system to a more secure 8-bit equivalent. The enhanced performance is largely due to the power of neural networks, which have grown from just five layers in 2012 to over 1,200 layers today. These deep networks are trained on massive datasets to extract meaningful patterns, which are then used to improve smaller, more efficient models deployed in real-world applications.
Wang explained that while front-end devices may use simpler models, they rely on complex, deep networks in the background. He likened this process to a primary school student learning from a highly skilled teacher. The teacher—representing the 1,200-layer network—absorbs vast amounts of knowledge from large datasets and then transfers that expertise to the student, enabling it to perform tasks effectively with minimal input.
This algorithmic progress has led to numerous practical applications, making cities safer and more efficient. For instance, dynamic face recognition systems have successfully tracked criminals who changed their identities years ago, and helped locate missing elderly individuals or children using city-wide camera networks.
According to Shangtang’s research, the accuracy of face recognition has steadily improved over the years. In 2014, training on 200,000 faces achieved a 98.5% accuracy rate, while human accuracy was around 97.5%. By 2015, using 300,000 faces, the accuracy reached 99.55%. In 2016, with 60 million faces, the error rate dropped to one in a million, and by 2017, with 2 billion faces, the misrecognition rate was as low as one in 100 million—making it applicable across various industries.
On the hardware side, Wang also mentioned Shangtang’s collaboration with Qualcomm. He believes that for AI to become widely adopted, it must be integrated into front-end devices. These devices depend heavily on specialized chips, and Shangtang’s face unlock technology is already used by over 100 million mobile phone users, relying on Qualcomm’s chip support.
September 27, 2025