Big Idea Challenge: "What would be the next big things in Computer Science?" - Submitted Ideas

Voting for Finalists

Voting Period: March 4 (Friday) - March 15 (Tuesday)
Voter Eligibility: SENG undergraduate students
Note:

  • Each eligible voter has the right to cast at most 5 votes
  • Voters award: All voters will have a chance to win HK$100 Starbucks coupon (20 winners, lucky draw among all students who vote)

Which one is your favorite idea? Let's vote for the finalists!

Voting closed.

Idea List

Idea 01
Quantum Computing - Managing Big Data in a palm of the hand

As Peter Sondergaard once said, "Information is the oil of the 21st century, and analytics is the combustion engine”.

Truly, we are living in an overflowing amount of data. According to Statista, there are 79 zettabytes of data generated worldwide by 2021.

As the amount of the data overflow, those are used for multiple purposes including predictive analysis, pattern recognition, consumer behavior, optimization, and more. But what Big Data can achieve the most in the future is in implementing the quantitative approach in the financial industry. What is considered unstructured data (communication record, news, qualitative reports…) are now systematically accumulated and modified to be trained in deep learning models.

Along with the explosive speed of its growth, however, there are more and more needs to handle such an overflowing amount of data. That is where quantum computing kicks in. Quantum computing is ​​a field in computer science that allows calculation performance to be based on a quantum properties such as superposition, interference, and entanglement. Under extremely cold temperatures, the superconducting chip takes the quantum properties. In such a state, what bits traditionally operated by 0/1, qubits have both the 0 and 1 state. When explained in analogy - if the classic computer is a mouse iteratively trying to escape the complex maze, quantum computing is a mouse experimenting with all possible combinations all at once. The advantage of such a state is its computational power. Qubits allow tunneling because it has both wavelength properties and particle properties. In that case, only 20 superpositioned qubits allow 2^20, or 1,048,576 configurations all at once.

Although there still are technical limitations in implementing in real-world problems, Big Data with the higher computational power of quantum computing will ultimately change the paradigm of the financial industry.

Back to idea list
Idea 02
Headband Communication Aid for Alzheimer's Disease Patients

With an aging world population in which the total amount of people over 60 years old could double from roughly 1 to 2 billion by 2050, the number of individuals diagnosed with Alzheimer’s Disease may also increase. As patients with Alzheimer’s oftentimes start finding it more difficult to express their thoughts or needs due to decreased cognitive abilities, it would be useful to develop something to aid their ability to still communicate effectively through other means.
An idea is to develop a headband which could read an Alzheimer’s patient’s brainwaves and directly return (on an inbuilt monitor on the headband) using simple phrases or terms what the patient may be trying to communicate. For example, if the patient wants to go outside, the headband may be able to discern the patient’s brainwave patterns and return the phrase “wants to go outside” to signal that message to others. Implementing such an idea at present may still be difficult, but it is not impossible. It would require advanced Unsupervised Learning algorithms and also neural networks to be trained to 100% accuracy. Essentially the algorithms in the headband would slowly start learning from experience by identifying certain brainwave patterns as associated to respective certain ideas, words, or phrases. Then, it would be able to classify the patterns to start returning the phrases or words on the monitor associated with that pattern to aid communication.

Back to idea list
Idea 03
The Next Generation of Mixed Reality

My idea is that we link a stereo camera with a 3D projector/hologram system (multiple different types, but I am currently thinking of the fan like projector system). How will this work? The stereo camera will observe the user's movements, and based on that it will change the projection model shown by the 3D projector/hologram. Why is this useful? Well our current mixed reality/ virtual reality systems require the use of either virtual reality goggles or smartphones. However, I feel that this creates in inability of synchronisation with our real world actions. Thus, the advantage of this system is that we as people will be able to have an interactive system that feels like its a part of our own reality. Furthermore, injuries caused by virtual reality systems can also be alleviated. I see hope for this technology to one day replace powerpoint presentations. This is because, the inehernt point of presentations is to get your idea across in a visual manner, and what better visual manner is there than an interactive system that does not require you to wear/use any extra technological equipment.

Back to idea list
Idea 04
Quantum Turing Machine : From Microscopic Research to Cosmic Calculations

I believe that the Quantum Turing Machine will be the next big thing in Computer Science. An abstract model and design for the system exists. However, I believe that future development of society necessitates Quantum Turing Machines to be materialised as physical Computer Systems. Quantum Turing Machine will allow us to perform quantum computing, which in turn will allow harnessing the properties of quantum states and using them for advanced calculations: quantum interference and superposition. As, we continue to delve deeper in the world of particle physics to discover new forms of generating energy, such calculations will be a must. Quantum Computing can also solve existing computational problems such as Integer Factorisation and RSA Encryption significantly faster. This is important since cybersecurity in the digital space is gaining increasing traction. On a macroscopic level, the quantum computer will allow us to perform advanced calculations - those of cosmic proportions - such as the current state of the universe, which would not be possible using a regular computer.

The Quantum Turing Machine can be designed using various quantum logic gates, which use advanced boolean logical units known as qubits. To materialise the qubits, we can use transmons and ion traps. This will allow them to function as regular bits with 0 and 1 quantum states. In order to stabilise the quantum states in the system, the Quantum Turing Machine will be able to perform error correction by employing redundancy and ensuring decoherence of noise. With the Quantum Turing Machine being the next big thing in technology, current quantum theorems and algorithms can be used to reduce the time complexity in standard algorithms, thereby allowing for more efficient Big Data Processing.

Back to idea list
Idea 05
借一下(JYX - Jie Yi Xia) - Rental Community Platform

借一下(JYX) is a rental service platform with a delivery service. It is a community platform that works in both web and app. The service can be explained from two perspectives: renter and rentee. For the renter, the services let them register any item they want to rent to others with the price they think is reasonable. For the rentee, they would search for the items they would like to rent, and if they find the item with the price range they want, they can contact the renter through our chat system. After the renter and rentee discuss the price, contract, and other specifications, they can sign the contract through the app. After the rental contract has been signed, the renter could deliver the item themselves with less commission fee or deliver through our delivery service with more commission fee. Whenever the renter decides which way they would like to deliver the item, the money would be automatically transmitted from the rentee to the renter.
The revenue model of this platform is the commission. Whenever the transaction happens, there will be a certain percentage of the commission fee taken from the transaction and the remainder would be given to the renter. If this is not enough, there could be an additional commission fee for the renter to pay whenever they register the item they want to rent. If there is a company that wants to register multiple amounts of items, they can sign a contract with us and pay all commission at once, but at discounted price.

Back to idea list
Idea 06
Mind-controlled humanoid

The human body has many limitations. However, those limitations are what define us as a human at the same time. In the real world, there are many accidents that can cause irreparable damage to our bodies. It will be convenient to have a second human body that we can have full control of, can accept whatever risks that may damage our original body, and can act exactly like our original human body. A solution to give us a second human body is a Mind-controlled humanoid. With the humanoid, we will be able to participate in high-risk activities, such as climbing Mount Everest, without risking our life.

Compare with those human body simulations in VR or AR, which help humans to explore the world virtually or even experience fantasy stories, the aim of the Mind-controlled humanoid is to complete tasks in real life with the real “second body”. For instance, in the future, there could be Mind-controlled humanoids working in construction sites. If the humanoid unfortunately died because of an accident, the person behind it doesn’t actually die.

For feasibility, there are several currently developing technologies that can contribute to the development of Mind-controlled humanoids. First, network speed technology is evolving at an exponential rate. The current 5G technology has low latency and is capable of remote surgery [1]. It is clear that in the future, the network technology will be capable of more precise human actions with extremely low latency. Second, a company called Neuralink is developing a brain chip device that can capture brain activities and turn them into instructions [2]. This kind of technology can definitely be used in controlling different kinds of robots. With similar technology, the control for the Mind-controlled humanoid can be built.

References:

[1]
https://www.cnet.com/tech/mobile/5g-not-just-speed-fifth-generation-wireless-tech-lets-you-do-vr-self-driving-cars-drones-remote/

[2]
https://www.businessinsider.com/neuralink-elon-musk-microchips-brains-ai-2021-2

Back to idea list
Idea 07
Machine Learning Architectures to Better Approximate Human Brain Neural Networks

Certain assumptions are always made, and specific rules are always followed when designing a new neural network architecture — the weights inside the neural network do not change during the processing of a token & internal weights do not change during testing.

Both of those make it harder to approximate the working of human brain neural networks as recent research shows that connections between neurons undergo continuous change during information processing.

Also, recent research shows that the electromagnetic field created by the brain, which EEG can measure, is not a side effect of brain function. Instead, it can affect the "weights" of axons (links) between neurons across the brain.

As a result, creating better models of human brain neural networks to design better natural language, computer vision, reinforcement learning algorithms, the weights inside the networks should be allowed to change continuously.

I propose that the output of each neuron is input into a function that outputs a matrix of adjustments to all weights in the network. The adjustments are applied after each layer of the network. The function is learned during training and does not change during operation.

My proposed architecture has a high probability of enabling better computer vision, natural language and reinforcement learning algorithms. It allows for a more complete text representation, possibly enabling better question answering algorithms. The video processing algorithm could potentially be improved as well. Improvement in machine learning will enable automatization of some data processing jobs and reallocation of societal human brain capacity to other tasks for more effective utilization of human resources resulting in improved productivity (as defined by the macroeconomics).

Back to idea list
Idea 08
AI Personal Assistant

AI Personal Assistant is a smart assistant that helps us with every daily decision, what makes it extraordinary among today's chatbot or AI is that it is completely personalised.

AI Personal Assistant should exist as a chip inside our body, or as an application on mobile devices like phones and computers. To initiate our own AI assistant, we have to pass in enough data in various forms, such as our messages, social media pages, photos and so on to understand our personal preference. Based on these data and some inputed background information of ourselves, AI assistant would help us to analyze and make decisions according to our preference and the pros and cons of the options.

I believe personal assistant would be the next big thing because nowadays information and options are flooding us. When we go online shopping and look for the best outfit that suits certain events, I may get hundreds of thousands of results. We may try to go over them at first, but at last we always give up and pick the "best" in the first 100 results. There are countless similar situations in our lives and AI Personal Assistant could help us filter the options, saving us lots of time but getting a better result.

Back to idea list
Idea 09
Ropet - Your Mobile Backpack and more

Ropet is your personal assistant and mobile backpack. Just like your cats and dogs, Ropet always stays with you. It walks itself, and it can follow you. It even helps you carry your stuff like your phone, keys and water bottle so that you can free your hands and shoulders. The AI inside Ropet also gave you different kinds of assistance. It can tell you the date, time, and weather, just like the AI inside your phone did. But what's more? It can help you intervene in the physical world, which your phone cannot! "Ropet, give me my phone." Not only can Ropet give you your stuff inside it, but it can also take things somewhere else for you. "Ropet, take me my pills over there," "Ropet, grab my keys and follow me". Doesn't it sound convenient for you?

Compared with traditional backpacks, why Ropet? It is super secure. Ropet never gives your things to others without your permission. No one dares to steal things from you. The eyes of Ropet take pictures of any thief who is forcing Ropet away from you. It can upload the photos online and call the police itself. If someone really stole it anyway, the GPS inside Ropet ensures your stuff would never be lost. (Much better than AirTag, right?) That's why you can also call it a mobile safe. You can also download different modules to make Ropet even more multi-functional. You can let your Ropet play different games with you.
(P.S. With strong enough modules downloaded, Ropet can even take a bullet for you)

Back to idea list
Idea 10
self-constructing program

Self-constructing program, can identify problems by itself, and edit the code by itself. Hence, it is a problem solver for a dedicated problem with a few lines of primitive code. By doing so, programmers do not need to work on highly repetitive jobs. Many products can be done with it, such as personal assistants. With the power of AI, it allows the program to understand the problem(such as NLP), so as to optimize or add function to itself.

Back to idea list
Idea 11
Content Personified: Search Less, Enjoy More

You're sitting in a bus and staring out the window when suddenly a song pops into your head. You focus in on the tune repeating over and over and the fragmented lyrics that come and go but the name is no where to be found. Being stumped like this bugs you. Why does recovering a lost song from your past always have to be detective work? Why can't your technology provide you this convenience?

If such a feature existed then media streaming platforms as a whole would work better and align themselves with what the user wants in the moment. Currently these platforms provide recommendations solely based on what the user knows they want. However, just as the iPad was something users did not know they wanted/needed, users on streaming platforms often don't know what they want or at least don't find it easily accessible to them. I think we can all relate to finding ourselves scrolling and searching more than enjoying the content available on a platform.

Combining an EEG layered cap or EEG layered headphone headband, heart rate readings from an Optical Heart sensor, skin conductance from an Electrodermal Activity sensor and possibly heat signature from a Thermistor (the latter 3 could be in a smartwatch) the song/movie/topic/book/article/paper that is in the mind of the user and the estimation of the emotional reaction the user has when thinking about the said item could be used together with a neural search engine to make the content appear on the app.

The hardware (sensor system) and software (search engine) can be built in parallel and confirm the progress and accuracy of the other simultaneously. The search engine will be trained to take in sensor data and an ML/DL model will be built to provide a shortlist of targeted and accurate content for users to either rediscover afresh or discover for the first time.

Back to idea list
Idea 12
Contactless Charging

I am in absolute love with Lithium ion batteries, I think they are one of the most brilliant recent inventions. However, a downside I face with them is that they do not age well. Every couple of years my phone battery deteriorates significantly and I seldom find myself in situations to have enough charge for the rest of the day. I wish we had wireless chargers installed in rooms where I can just enter a room and the batteries on my devices charge automatically. Imaging entering your room and you do not have to plug your phone into any charger. When you leave a room, you can be almost certain that your phone will be fully charged. Due to this, I think complete contactless charging can be the next big thing in computer science and electronic engineering.

Back to idea list
Idea 13
Carbon Exchange 碳交易所

My idea is using modern technology to change human behaviour by the Carbon Exchnage.

An average of four tonnes of carbon dioxide is generated by one person annually. In Hong Kong, a total of 40.1 million tonnes of carbon dioxide were produced in 2019. When the Kyoto Protocol was enacted, countries and companies alike have adopted both Voluntary Emission Rights (VER) markets as well as Compliance Emission Rights (CER) markets to meet pre-set milestones, with the additional benefit of increasing corporate social responsibility and gaining recognition. This shift in the landscape gives rise to the emergence of the carbon offset/trading market. 

In China, the National Development and Reform Commission (NDRC | 國家發改委) announced that corporations in the following eight sectors, namely, petrochemicals, chemicals, building materials, steel, ferrous metals, paper-making, power generation, and aviation, would be included in these markets. The aforesaid companies with more than 10,000 tonnes of carbon emission per year are compulsory to participate in these markets and are required to use a more designated amount of energy. By 2021, the carbon trading price ranged from RMB40 to RMB90 per ton, and soon the right to emit carbon will become a scarce resource. It is forecasted that there will be a huge potential in the carbon trading market in the future and individuals can participate in the ecosystem of low-carbon life in the future.

My idea aims to establish this vast untapped carbon offset market. Developing a platform to spearhead C2C carbon offset to the public, providing clients with incentives to offset their carbon footprint through daily tasks calculated into negative carbon emissions. The pooled carbon quota acquired is sold back to existing cap-and-trade emission trading systems (ETS). Differentiating ourselves with the traditional B2B or country to country cap-and-trade markets.

Computer Science could build a brand new sustainable economic ecosystem, and popular worldwide carbon offset collection and C2B carbon trading platform to save our mother nature!

Back to idea list
Idea 14
Brain-Computer Interface and Wireless Connection for Phone

Imagine that you can use your phone, open any app, type texts without touching screen anymore, all you need to do is to think the name of the app or content of the texts. To search something, just call google in your mind and a small window will pop up. To play a music, it will be even more convenient since everything can run background. Even no need to slide the page when lying in the quilt, the phone can be supported by the bracket and no need to move the hand out of the quilt anymore.

Back to idea list
Idea 15
Crime and Accident Blackspot Map

My big idea is about Crime and Accident Blackspot Map, since many Hong Kong residents are world travellers, we like seeing different places with friends or alone. However, because we cannot know how that area is safe enough before we arrive, several investigations must be conducted during the trip preparation process. Therefore, a Crime and Accident Blackspot Map is proposed. This map will feature basic map functions, but one of the most notable differences from the present map is that it will indicate the crime rate in various districts, with an emphasis on robbery, thievery, home break-ins, gun violence, and other crimes involves violence. This tool allows users to be aware of the potential for crime while planning a visit and to take precautions beforehand. Furthermore, accident blackspot will be cautioned and highlighted to the user, assisting the user in being cautious of those locations, particularly for those who plan to drive. These may be done by employing data mining techniques to calculate the crime rate and locate the most popular blackspot based on past data, then linking them to the GPS and providing real-time feedback to the user through sound, pop-up notice, or vibration.

This application may assist not only the typical user in ensuring their own safety but also the government and police in reducing and preventing crimes and criminal disorders. This tool is suitable for practically everyone who has a smart device, and the long-term benefits include reducing crime, reducing the occurrence of accidents, protecting people's property and personal safety, and so making society more affluent and stable.

Back to idea list
Idea 16
Dream manipulator

Sleeping is one of the most important activity to human. In average, people spent one third of their life on sleeping. Dreaming is the only thing we can do when we fall asleep. What if we can take control of the dream and save the contents of it into a computer for further use? We need a dream manipulator to utilize the power of dream.

The dream manipulator allows us to output thoughts to the machine. We can use our imagination and create something new while asleep. There are examples of people being inspired by dreams. Inversely, the computer can provide scenes for the dream. We may travel around the world and have a relaxing holiday without leaving our bed. With such device, individuals can be financially benefited from high productivity as they can still work while sleeping. On the other hand, people can stay away from nightmares and enjoy a restful night. In a broader perspective, the development of economic, as well as the technology, would be significantly boosted under the increasing productivity and creativity, especially when such devices are used to help generating inspiration.

Dream manipulator could be a feasible idea as there exists successful experiments of directing and recording the dream. For the dream saving part, the computer needs to extract data from our brain through electroencephalography or other methods. Then, the data would be processed into forms of multimedia via AI and stored in the computer for users to review. For the part of directing dream, we would need devices that generate sounds or smells during the state of hypnagogia to assist user to dream of their desire scene. Last but not least, the key of the whole invention is to create lucid dream, which users know they are dreaming. This would be the basis for user to control their dream.

Back to idea list
Idea 17
IoT-optimized Time-series DBMS

At first glance, the future development of the market is very likely to be headed to high-end technology like artificial intelligence and blockchain. It is indisputably effective when it comes to replacing repetitive and uncreative jobs like bus drivers. However, I believe the future would demand environmental technology more than ever. There is a surge in “Environmental, Social, Governance (ESG)” related investment in recent years. Sustainable bonds have grown 76% year on the first half of 2021 and the market size of environmental technology is up to 550 billion USD. People are more aware of their impact on the environment and put more emphasis on it. It is because all the technologies we looked upon are still powered by electricity. And nature is our fundamental source of energy. As Elon Musk mentioned that Tesla would accept Bitcoin when the energy of mining them is cleaner, leading tech and finance companies could focus more on the integration of technology with sustainability.

The big idea I am proposing is a time-series database management system specialized in controlling IoT sensors and devices. The database would collect usage and sensor data regularly. Over a prolonged period, we can use the program to map out and identify areas of inefficient energy consumption. It could give out patterns and forecasts of possible ways to use electrical devices. Take HKUST as an example, the corridor lights in UG Hall are turned on 24-7 where the motion sensors are likely to detect very few signals at midnight. We can easily identify the waste of energy in that period and the location. This idea also requires the replacement of old electrical appliances with ones integrated with internet connections. The database could be written in C++ with a C++ API and SQL data access.

Back to idea list
Idea 18
Technology for the blind

Our smartphones have developed from Nokia and Blackberry headsets, a bulky, inefficient processing device to smartphones, a light, faster processing machines. However, a large segment of our society has been constantly neglected from this development: the visually impaired. With new features coming in to our latest smartphones, they neccesitate the need for visual cognition, which the visually impaired cannot enjoy.

Hence, I propose the solution of a Refreshable Braille Screen called Brailliant, which is an adaptive braille screen that can be attached on top of any smartphone, and will convert the text and images into braille on the screen. Since, it is adaptive and refreshable, it will allow users to interact with the smartphone and access apps and their functionality. Further the braille screen can also produce images, something that the visually impaired could never interact with. here are 17,500 visually impaired individuals in HK that are economically active (10.1% of visually impaired population). Among this economically active population, close to 13,00 individuals (75%) have a monthly income of more than $7,000 HKD. I believe with this product they can access digital literacy and really uplift their lives and society.

Back to idea list
Idea 19
Adapting Permissioned Blockchain to Management System

Nowadays, permissioned blockchain has been well developed and introduced to the public, such as Corda or Hyperledger Fabric. Permissioned blockchain can provide integrity, confidentiality, and scalability for data on chain, while does not require any miner or Proof of Work process. Its functionalities can be defined by smart contract, its content can be shared or verified across different types of blockchain by protocols like weaver. Security of such system depends on public key infrastructure where Certificate Authority provide access to other members of the organization. Peers of such network can be expanded to form different tiers of permissioned blockchain.

This can help streamline governments and enterprises and eliminate bureaucracy. Permissioned blockchain can be formed by peers in departments of different levels and share data to people who needs to know. Verification can be done across different permissioned blockchain of governments or companies without knowing plaintexts. Authentication of such system can be guaranteed by digital signature issued by the Certificate Authority of the blockchain, such as the senior members or officials that can be trusted. This system does not rely on single server at one place, and thus are more robust against unstable network. It is also more transparent and efficient to the public since it also provides ways for the public to verify and monitor when certain information is needed to be known. Thus, I believe such technology could reform the existing centralized managing system.

Back to idea list
Idea 20
Mobile Blockchain as a Distributed Supercomputer

My big idea is that edge computing devices, such as mobile phones and internet-of-thing devices can form a giant blockchain that utilizes idle processing power to process scientific data. Blockchain technologies can create an immutable public ledger that is immensely mathematically difficult to forge. However, the calculation done for a blockchain are only useful to verify the validity of transactions without contributing to scientific discoveries. This makes each blockchain network incurs a large power usage and great environmental impact. More advanced cryptocurrencies try to mitigate the environmental implications of this property by implementing proof-of-stake over the more common proof-of-work, but the migration has not been complete.

A blockchain targeting mobile devices can be formed to tackle scientific data processing. First, modern mobile devices have amazing processing power. Back in 2014, mobile phone manufacturer HTC created the “POWER TO GIVE™” program where users can install an app and uses their mobile phone to contribute to a distributed supercomputer that helps scientists process large datasets. Almost a decade have pasted and our mobile devices have gotten more energy efficient with application specific hardware for acceleration. Second, idle power usage can be comparable to a throttled load on these mobile devices. With so many active mobile computing devices in the world, the accumulative processing power can rival the top 500 computer list at a minimal energy cost. Third, validity of the computation results can be ensured by the blockchain. Hashing difficulty can remain low since majority of the processing power are dedicated to data processing.

With rewards directly comes from the scientific institutions publishing the tasks, this blockchain supercomputer has the environmental and financial bases for it to be self-sustainable, and I believe the social acceptance of using idle processing power for good can gain traction sooner rather than later.

Back to idea list
Idea 21
Deep Learning in Advanced Reality

With the knowledge of machine learning, we are now capable of using it to train many models to simulate many human behaviors. By applying this tool to deep-learn the environment aroud users, the computer will have a 3D model of the surroundings. Thus, the disadvantage of current AR devices or applications, known as slow response to the change of views and environement, can be solved.

By doing so, accurate volumn densities and colors can be recognized by the trained model and rendered quickly. And the application prospects of this technology is broad. Virtual presentation, remote guidance, simplify the devices and so on. With the help of high speed 5G communication, the AR can be used in many unexpected new areas.

Back to idea list
Idea 22
Artificial Intelligence Analyzing Sound Frequency and Music

Any sound frequency can be presented in a form of a spectrogram for one to analyze multiple frequency components that coexist in complex patterns. The idea of this proposal is to use artificial intelligence to gather, analyze, predict, and utilize the uniqueness of sound waves.
Analyzing sounds can be seen as extraction from an audio signal sound source such as amplitude, distortion, and frequency response. To elaborate, prospective usages of sounds waves are music therapy, restoring voice changers, sound print locks, etc. Restoring voice changers assists in criminal investigation, and sound print locks can be utilized in future technologies to control virtual electronics or open digital locks or give commands by a specific owner. Uniqueness in the sound frequencies is the key.
Yet, one might wonder why music therapy. As its name suggests, music therapy is the clinical use of music to accomplish individualized goals such as reducing stress, improving mood and self-expression. Yet, qualification as a therapist in this field is extremely hard in terms of proficiency in psychology, music theory, as well as the composition of songs. With the aid of sound frequency via artificial intelligence, the latter can be realized through the interpretation of chords and notes in a song by a computer. By linking to the emotions that brought along one has while listening, any composition can be generated with analyzed data (specific chords and notes), and it alleviates the requirements of a psychologist required to qualify for this position. Not only patients inside the medical room can be taken care of, but these music generators can also be healing to people who seek music meditation.
Overall, analysis of sound frequencies and utilization is underestimated nowadays. The prospect of sound waves can truly be the solution to mental consolation as well as the technological tools humans are seeking for.

Back to idea list
Idea 23
Biochips and Nanorobotics

Biochips are essentially made up of a core microchip and an array of molecular sensors on the chip surface, which should be able to analyze various biological elements and further transmit the data back to the computer. Well-rounded algorithms are needed to connect the biological sensors and the chips as well as post back the statistics without any delay. Hopefully soon, this kind of chips can be tiny enough to implant into the patients’ body for real-time monitoring. The chip can provide timely feedbacks to the computer in case any component changes occur within the patients’ body. The successful implementation is likely to gain insights on controlling incurable diseases like cancer. This big idea will undoubtedly revolutionize both the Computer Science and Medical Science industries.

For those wanting to go even smaller than Biochips, Nanorobotics should be the following big idea. Under the control of a central computer, Nanorobotics could potentially be used to identify and destroy hazardous cells (e.g., cancer cells), measure the concentration of certain toxic chemicals and deliver drugs to a targeted position. As the overall procedures tend take place within the human body, a strong and secure set of algorithms should be designed to minimum the potential risks. Therefore, countless bug fixes and experiments should be done before deploying to production. If the Nanorobotics can win wider appeal and be accepted by the general public, it will also become a milestone in Artificial Intelligence field.

Back to idea list
Idea 24
eMentor

An artificial intelligence-based application that helps beginners in any problem-based field, for example, coding who despite studying/understanding the basics are struggling to solve the problems by guiding them to the solution through vocal interaction. Just as the famous ‘rubber duck’ among programmers, so instead of explaining the code to the duck and struggling to find the solutions all by themselves, the programmers would instead talk to the AI assistant through the application and it would guide them through their codes by interaction and hinting towards different possible scenarios the programmer can opt for, without giving them a straightforward answer, which will help the programmer arrive at the solution by themselves and better understand the problem.

Similarly, it can be expanded to different fields like Math, cooking, painting, etc. The person would use the virtual assistant as a guide to arrive at the solutions to their problems, by hints provided by the assistant similar to how a teacher or a mentor would advise you without giving you direct solution to the questions.

This can be popular among people of different age groups who are willing to try their hand at something new and have no prior experience or guidance in that specific field. People who do not like just looking up solutions without any detailed explanation can also make great use of this application.

Back to idea list
Idea 25
AI Robot Teacher

Do you know teachers work 50 hours per week on average? Based on a survey by EdSurge, 50% of teachers have seriously considered quitting in recent years. There is a need to reduce the burden of teachers, therefore introducing an AI robot teacher that will handle all the duties of a normal teacher. Through designing the course structure, the robot teacher can base on the students’ ability to plan out the whole term schedule. By analysing the students’ behaviour and character, it could also recommend the teaching style and the activities to be held in class. On top of that, it could tailor-made assignments and exams based on the student’s strengths and weaknesses, so to enhance the students’ learning experience. Due to the digital transformation of the world, many products like cars have been automated. The chores that teachers have may also be automated by AI. In terms of social aspect, it could not only reduce the burden of teachers, staffs and assistants, but they could also focus on the quality of teaching to the next generation. Furthermore, in this data-driven world, the marks of the students’ assignments and exam is useful to determine their weakness, so to provide additional exercise. By providing a better education to the youths qualitatively and quantitatively, they could even outshine the previous generation. Several technologies will be used in this idea. Robotics will be used for building the robot teacher. While computer vision is used to scan the students’ assignment and analyse the students’ in-class behaviour, natural language processing is used to mark their assignments and exams. Big data is stored for the teaching resources, and finally machine learning such as reinforcement learning and deep learning is used to tailor-made the student learning process.

Back to idea list
Idea 26
AR as the new means of the socialization among people

In the past two decades, we have witnessed a significant development in communication among people. From the earliest email to instant messaging and real-time video chat, the means for social has been advancing at an unprecedented speed and consequently introducing new business models that shape people’s lifestyles tremendously. With the mature hardware technology, Augmented Reality (AR) could be the next generation of social means. People may use AR to better interact with each other in a more immersive and engaging manner.
Specifically, the personalized AR avatar preloaded with personal information such as hobbies could be a new form of communication comparable to the account on other social platforms. In this way, the AR avatars may interact on behalf of the creators based on their expectations, helping the creators to make the first step of the conversation, which should be highly beneficial to those with social anxiety or phobia. For example, in an icebreaking scenario among a small group of people, each of them may create their AR avatars and let those avatars initiate or facilitate the conversation based on their similarities or difference. Or, in a larger event, such as a career fair, the AR avatars created by individuals may increase the chances of exploring companies with more ideal matches for them. Financially speaking, each leap of the transformation of the social means would make a tech giant rise. Facebook raised by introducing instant messaging. ByteDance, the holding company of TikTok, raised by introducing the short video streaming service. The transformation to AR socialization might as well make a new giant grow, especially with the emergence of AR glasses.

Back to idea list
Idea 27
Delivery X Drone (Dronlivery)

Is drone delivery possible? In my perspective, it is definitely a yes in the foreseeable future. My big idea of delivering food or goods by drone is inspired by my own work experience, future trends, as well as Industrial IoT technology.

First and foremost, being a Foodpanda walker for one year, I see food delivery in high demand. Almost, if not all the students in HKUST knew and used food delivery apps such as Foodpanda and Deliveroo because of their convenience and huge food variety. Moreover, in the view of the COVID-19 situation, our society has developed a “Work From Home” atmosphere and people are less likely to go out for lunch or dinner. Furthermore, Virtual Reality and metaverse would be the next-generation technology, community can be built in the network and no more border barriers. Once the general public is prone to networking in Virtual Reality, drone delivery is definitely essential for sustaining life and bringing great convenience.

For the plan of drone delivery, first, the normal drone would have a low battery life, the flight time of most high-quality drones is about 20 minutes. With weight, the flight time of the drone would be much shorter. To deal with, a battery with a larger capacity can be installed inside the drone regardless of the drone size and quick charging after one delivery by setting more recharging regions in different offices around the city. Second, the bearing capacity of the normal drone is less than 5 pounds. However, by using a powerful turbo-generator, the generated lifting force can be much higher. Third, concerning location detection and difficulties in delivering to the specified floor, IoT technology such as GPS autopilot and machine learning allows machine-to-machine communication between drone and phone as well as accessing huge amounts of data in the cloud to gather more specified information. Fourth, drones can use an isothermic bag similar to normal food delivery to ensure food safety. Fifth, privacy in drone traveling is controversial. In order to prevent violating residents’ privacy, drones could install a LiDAR camera instead of live camera to detect the surrounding 3D image.

The roadmap is that we first R&D in Hong Kong, then Taiwan and Japan become the first countries to implement drone delivery systems due to their low population density and food delivery culture, thus implementing in Hong Kong and South Korea, and spread worldwide.

If a city can implement an automatic drone delivery, it would have a great influence on traditional delivery systems as well as the lifestyle of the citizens. If drone delivery is well developed, drone merchandise and systems can spread worldwide and the market size is unprecedented.

To conclude, drone delivery is feasible and of paramount importance in the foreseeable future. As our technology becomes more and more advanced, this big idea should and will implement in the future.

Back to idea list
Idea 28
Expand the idea of GitHub in Video Editing for cloud-based cooperation

In recent years, as online video sharing and social media platforms, such as Youtube, TikTok and Instagram become popular, the video editing industry grows exponentially. However, according to a friend of mine who works in the industry complains, there are a few issues that lead to inefficiency in the cooperation of video editing teams. His team suffers from an unbalanced workload because they cannot work parallelly. And he needs to export several versions with different fonts for his clients to pick one. He also had to invest a lot in desktop computers with powerful GPUs.
As a CSE student, I immediately came up with the idea of migrating Git and Github in the video editing field to make a version control system and a cloud-based hosting server.
With these, the editors, special effect designers, and animators can make their own changes in a single video source separately and the version control system will record their changes, such as adding masks and video trimming. The changes should be recorded as operation instead of differences between the current version and the previous version because unlike the text field such as code, the video trimming will definitely cause conflicts in merging branches, and recording the changes in operation can handle most of the conflicts. The editors can push their local changes into the cloud. With the help of cloud computing, the hosting server will store the video sources and apply the changes (operations) they pushed one by one, then render and export to publish on the webpage. Editors can have pull requests and the team managers can review their branches and merged them into the main branch to complete the video editing task. Not only can this cloud-based hosting server save time for the video editing team by enabling multithreading, but it can also make online video editing possible. The editors can save their money for upgrading their computer configurations because they can work in any local devices connecting to cloud servers by sending the operations to the cloud, just like the software developers coding in remote servers by ssh. Their current process and semi-finished videos can be deployed on the webpage and shown to their clients, which will save time and resources for communication. Clients can open issues on the web page to specify their needs and ask the team to modify them.
This version control system and the cloud-based hosting server will solve these three issues I mentioned above.
The latest research in Feb 2022 found that the video editing software market size was valued at $1.94 Bn in 2019 and is projected to reach $3.04 Bn by 2027. My idea of a version control system can serve as a plug-in in current video editing software, and the cloud-based hosting server will change the current mode and create a new market. This system will improve efficiency and effectiveness in the entire video editing industry, so I believe this project is profitable.

Back to idea list
Idea 29
Metaverse

Metaverse is a buzzword used to describe 3D virtual spaces for social connection or recreational purposes. While many companies, including Amazon, Microsoft and Meta, are competing in their VR world development, none of them has managed to create a full-encompassing experience that blurs the line between reality and imagination. A metaverse in its fully realized form offers you true-to-life sights, sounds and smells. Any at-home traveler can breathe the air of ancient Greece or enjoy the taste of a fresh warm strawberry that just fell on loamy soil, simply with their AR/VR headsets or smart clothing worn.

A novel way of social interactions, a gateway to another means of living and sensory experience, Metaverse could be as authentic as any other physical interaction. As the ecosystem of shared virtual worlds develops, the financial and social impact throughout its gradual development is immeasurable. Companies will need to transition their strategies from online ad buys to existing in a shared, virtual economy. Real-world celebrities will start to play games, and players will become celebrities themselves. As for the people, they will build up relationships with other users, or even NPCs and holograms.

Increasing attention in institutional and widespread interest, improvements in hardware performance and computational methods, collaboration among virtual world stakeholders and developers, are contributing factors to the formation of Metaverse. In addition, current technologies such as NFTs and blockchain play a vital role in the Metaverse. They help birth an open and fair economy that allows users and businesses to port real-world assets and services. Last but not least, it’s crucial to build standards and protocols to be widely adopted during its development.

Back to idea list
Idea 30
Golden Retriever

Have you ever heard an announcement of a missing child in a mall or lost anything in public places? Here is an AI system to help you and the panic parents. By providing a reference image or even verbal description, it applies object recognition to CCTV in the mall and locate the person or object instantly. Currently, to find a missing person relies on manpower and maybe announcement but there is security problem of announcing name of a child to anyone that increase the risk of abduction. Common way to find lost property is to check if it is returned to security centre or to revisit places where it might have been lost. Golden Retriever could enhance the efficiency and expand the coverage to places that you didn’t think of. It could not only increase the success rate of getting back the lost property, but also greatly reduce the manpower to search for an item or a person. If you have your CCTV installed at home, it could help you find out anything in less than a second, turning all your physical objects to an image reference and look for it in the CCTV recordings. For digitalized objects, it is much easier to perform a search and get the exact pieces of information even in this era of information explosion. With Golden Retriever, you can now press Ctrl-F to find them.

Back to idea list
Idea 31
DeFi - Decentralized Finance

Just a decade ago, fintech startups like PayPal and Robinhood outplayed traditional, clunky banks by taking a tech-first approach and offered users seamless access to financial services. However, the financial system is still far from perfect. Stocks, bonds, and other financial instruments take days to clear and require massive amounts of human capital. Billion-dollar baking scandals occur frequently behind closed doors of the privileged few. Banks hire thousands of employees just to keep maintaining inefficient processes and being compliant with ever-changing banking regulations. It is nearly impossible to start a new financial company without access to enormous amounts of capital. The financial industry is one of the least innovative in the 21st century, because of the rigidness of traditional banks, built on old and inefficient foundations.

Now comes DeFi (decentralized finance), leveraging the power of cryptography, decentralization, and blockchain to build a new financial system that can bring access to services such as payments, lending, borrowing, and trading, in a more efficient manner. All operations are settled almost immediately under the blockchain, and DeFi protocols can operate with little or no human involvement.

To explain just how “BIG” this idea is, consider the recent famous GameStop saga. Discovery of overly shorted stock by ‘powerful’ and ‘influential’ hedge funds was discovered by ‘weak’, ‘non-influential’ users of a Reddit group. They initiated a short squeeze that would result in complications for the hedge funds. Eventually, Robinhood and other stockbrokers came up with a controversial solution to disable the possibility of buying GME stock. In traditional markets where the powerful hold authority, a situation like this just wouldn’t be possible on a decentralized exchange.

There are just so many opportunities out there, and the future relies on DeFi. An example is the profitable DeFi business in the insurance industry. Coverage is bought from a decentralized pool of providers, instead of just buying from a single company. Anybody can act as a coverage provider and essentially, everyone is a liquidity provider, who will be exposed to risk, but will earn interest on the capital they lock up. With the help of Artificial Intelligence, faster and legit approvals will solve problems of insurance delays and help gain trust among the population. All in all, this is a rapidly booming industry and immediate action is required, as dozens of DeFi insurance companies have emerged around the globe just in the past year!

Back to idea list
Idea 32
Virtual Girlfriend

Intimate relationships are very important for mental health. However, at HKUST, due to the extremely skewed gender ratio of the student body, it is very difficult to find a romantic partner. Even outside HKUST, it is getting more and more difficult to seek romantic relationships, as social interactions become less and less frequent due too COVID.

The solution I propose is the Virtual Girlfriend. What!? Is it some kind of smartphone app where you open a picture of a 2D anime girl, and she says “I love you” or something? No! I am proposing something much more personalized and sophisticated.

The Virtual Girlfriend is not going in live in your smartphone. She will be 3D and real. As you may know, holographic technology has been increasingly employed in recent years. It can be used to project a moving image that looks 3D on a glass screen. It is the technology used to create the 3D image of virtual singer Hatsune Miku in her live concerts.

All that is needed is a plexiglass screen and several projectors. This technology is currently only used in concerts, but can be downsized to work in homes. Virtual Girlfriend will come with plexiglass screens and projectors that will work together to create a moving hologram of your Virtual Girlfriend that you can interact with in your home. Speech recognition and motion capture will be used to capture your speech and gestures, to which your Virtual Girlfriend will react.

Vocaloid technology will be used to allow the Virtual Girlfriend to talk (and even sing), using customized voices. The Virtual Girlfriend will come with an AI that will learn your daily routine and will react appropriately such as reminding you to eat, sleep and do your homework. The AI will also be able to carry out conversations. It will also keep track of your interactions. If you ignore your Virtual Girlfriend too much, she will leave you!

If this is going to be a commercial product, then how can be personalize each and every customer’s Virtual Girlfriend? Won’t everyone be dating the same Virtual Girlfriend? No! Using a neural network, we can generate every customer’s dream girlfriend, personalized in both appearance, voice and personality. The technology to generate artwork of 2D anime girls through a neural network has already been developed (see https://waifulabs.com/). On this website, users are presented with several artworks of 2D anime girls. They are asked to choose their favorite one, and this process is repeated several times. With the answers provided, the system generates a new artwork of a 2D anime girl from scratch based on the user’s choices. Once this technology is perfected, we will be able to generate Live2D, or even 3D anime girls based on user preference. This process can be extended to generate the personality of the Virtual Girlfriend. Thus, we will be able to provide each and every customer a Virtual Girlfriend that is like no other – unique and personalized for them in appearance, voice and personality. What’s better than a match made in the stars? A match made using a neural network, of course.

While the product is called Virtual Girlfriend, this product can of course be extended for the use of female customers who are looking for Virtual Boyfriends.

Back to idea list
Idea 33
Bazi Dating App

I want to make a dating app depending on BAZI(八字) which is an ancient Chinese metaphysics and I recently found out that it's quite accurate, it's not superstitious, and has many theories and evidence to prove it, it's more like an ancient big data.

There indeed is a big relationship between one's birth time and his/her life experience so want to make an app collecting people's data and using AI to analyze the data, as well as palmistry and Physiognomy or even MBTI, so can provide people with better life suggestion and better matching.

It's a good way to collect data for future research on prophecy and psychology. It would be useful and has enormous commercial value as I noticed that many companies nowadays secretly hire people based on this theory, especially finance companies, so using data collected from this app, we can provide human resources consultation for companies.

Back to idea list