Understanding Computer Vision with AI using Tensorflow

As the name suggests; the talk provides insights on deep learning; machine learning and AI using Google technologies as well as how we can enable complex IoT applications which would be shared by experts in the industry. The talk was targeted at various industries such as in autonomous vehicles as well as using Convolutional Networks for computer vision.

As part of Overflow, we were notified of the event by other students. As I was in the Project Empower AI lessons, I felt that this will be a good opportunity for me to better understand the current trends and progress in the industry. This was further encouraged by the lecturers from Project Empower. As the lesson crashes with the talk, I had initially decided not to attend. However, the lecturers said that it would be great if I can share the experience with the class. As such; me and another classmate; Yan Hwee registered ourselves just one week before the session begins.

Because me and Yan Hwee were there to gather notes from the lesson, I believe that we were able to link it back to what we were learning. Most of what was covered during Project Empower were related more towards data manipulation but this talk taught us several important concepts that we can take away with relations to various algorithms.

Firstly, we had a welcoming address by C K Vishwakarma; CEO of AllThingsConnected, and founder of IoTSG; talked to us on the relevance of the Internet of Things, along with the moral perceptions towards Artificial Intelligence. While we joked on how humans will be replaced by robots, he emphasized that it is only through working together where we will be able to fully utilise the benefits. After all, AI was modelled on the behaviour of nature.

Secondly, Sandeep who was the founder of CloudXLab talked to us on the concepts of Machine Learning as well as Neural Networks. I felt that his explanation was easy to understand. For example, he explained linear regression in terms of how similar it is to going to a new hotel and trying to adjust the heater so that it suits your preference. Through such examples, he tries to make it easier for users to learn machine learning through his classes.

Lastly, we had a talk by Aurelien Geron; an ex-Youtube employee; who gave us a look at the usage of You Only Look Once (YOLO) object detection as well as several other image recognition concepts like using Convolutional Neural Networks (CNN) to provide better results.

Aurelien also showcased several examples of how image recognition was used in Youtube algorithms as a trial to categorised different videos. What we found to be most frightening was the usage of a Generative Adversarial Networks (GAN) created by NVIDIA and used by an Uber software engineer to create Deep Fakes. Through this, we see the big potentials that AI has for our future.

What we learnt from the talk eventually went into our final project in Project Empower. Through the use of CNNs, we were able to increase the accuracy of our predictions. The talk also gave me a better understanding on how AI came to be as well as grasping knowledge on the multiple concepts. It provided insights within the industry as well as what the future holds for us.

Learning Journey to Alibaba Cloud (Singapore Office)

This learning journey was an educational trip with relation to our technopreneurship module. I would say that a few years back, it is one where we get to learn from a startup. However, the company is more mature now and there is definitely a different learning perspective to which we can learn from. However, it also feels like they were mainly promoting the services to which they offer.

There were two speakers whom were talking to us. There was a video showing the different applications to which Alibaba provides. Most of them were located in China showcasing the multiple advancements. One example is a fully automated smart shop with many features embedded into how it functions. Instead of disrupting the traditional shopping experience, most of Alibaba’s ideas revolves around the fusion of new ideas into existing one.

From this, we can see that many are able to benefit. Instead of wiping out traditional walk in shops; Alibaba brings their experience into such outlets; integrating the various technologies that they offer. We also see innovations that were made to suit this changes. For example, there are automated checkout in their Freshippo supermarkets. With robots being a common sight.

The 11.11 Global Shopping Festival also showcase how their ‘New Retail’ initiatives are successful in changing the shopping industry. While being entertained by various activities, certain products are introduced to promote viewers into buying them.

Another example are fashion shows where there is still a runway to showcase different wearables but through this, consumers can buy on their phone immediately if they like what they were seeing.

Amazingly, the company managed to move all their solutions and processes to the cloud. This meant that as long as you have a laptop and a connection to the network, you can work from anywhere providing convenience and a secure way of storage. Messaging is never easier through Alibaba’s messaging platform with many features and the ability to communicate within China itself. It has multiple features required for a business. For example, meeting features like video conference as well as tight integrations to Alibaba’s cloud services.

Additionally, we also got to know more of their cloud services and were briefed through the various features that are provided. This is useful to us as it gave us a comparison with other cloud services which were currently available. Some applications might include the training of AI.

Most interestingly, they had a unique work culture. Being in the company already takes close to 2 months of interview. 4 sets of interviews with 4 judges; and if anyone says no; you will not pass the interviewing stage. Through this, you can see how particular they are with their staff. They also do daily reflections in Chinese which might put off several interviewees. To add on, you are expected to be ready through the night as most people; some from all over the world; are only available at that time for meetups and conferences. This is definitely where the chat comes in handy.

Through all this, you can see the discipline that is required. Nonetheless, the staff enjoy their work and feel that there is a lot to which they can learn. We were also told not to be attached to our ideas and change wherever necessary.