Hyper.ai's latest updates this week

A quick overview of recent AI for Science developments

Addressing the mosaic integration problem of single-cell multi-omics data

The average screening interval for clinical applications was extended from 12 months to 31.97 months

IoU increased from 1.91% to 24.42%

Hyper.ai's latest updates this week

Contains collection download link

The four major new functions of intelligent scientific facilities are: large scientific models, generative simulation and inversion, high-throughput autonomous unmanned experiments, and large-scale trusted scientific research collaboration.

To live a decent life, there must be green workplaces

The diagnosis of ophthalmic diseases is highly dependent on image recognition, and ophthalmology is very suitable for the application of technologies such as deep learning. In order to further explore the potential value of deep learning in the diagnosis of fundus diseases, five ophthalmology centers across the country, led by Chen Youxin, director of the Department of Ophthalmology at Peking Union Medical College Hospital, worked with Beijing Zhiyuan Huitu Technology Co., Ltd. and Professor Li Xirong of the School of Information at Renmin University of China to jointly develop a deep learning system.

The system helped primary ophthalmologists improve their diagnostic consistency by about 12% and provided a new method for the automatic detection of 13 major fundus diseases. This article is a further interpretation and sharing of the study →

Hyper.ai's official website updates this week:

* High-quality public datasets: 8
1️⃣ AISHELL-1 open source Chinese speech database
2️⃣ DeepSymNet Deep Symbol Network Dataset
…

* AI4S paper cases: 2
1️⃣ Aiming at the world's most common cancer, Chinese scholars established the breast cancer prognostic scoring system MIRS
2️⃣ Shenzhen Institute of Advanced Technology of Chinese Academy of Sciences proposed SBeA, which analyzes animal social behavior based on a few-shot learning framework

* Popular encyclopedia entries: 8
1️⃣ Floating point operations per second FLOPS
2️⃣ Random Walk
…

For more resources, please see this article →

What is the meaning behind the social behavior of animals? Does the progression of reproduction, predation, defense, and establishment of social hierarchy reveal the process of human evolution? The precise quantification, identification, and behavior classification of animal social behavior can help reveal the role of brain function and mental disorders in the interaction process.

This research urgently needs to improve research efficiency and accuracy through technological innovation. Based on this, the Shenzhen Institute of Advanced Technology of the Chinese Academy of Sciences developed SBeA, which can comprehensively quantify the behavior of free-living animals, use a small number of labeled frames (about 400 frames) to estimate the three-dimensional posture of multiple animals, and use a two-way transfer learning strategy to achieve an accuracy rate of more than 90% in multi-animal identity recognition. This article is an interpretation and sharing of the experimental results👉

Breast cancer has officially replaced lung cancer as the world's number one cancer, with high complication rates and mortality rates, but if it can be detected early and treated according to best practices, the survival rate is expected to be greatly improved.

Recently, Chinese scholars used a neural network model to establish a scoring system MIRS for predicting breast cancer prognosis and treatment, which can be used to guide the formulation of treatment strategies for breast cancer patients. This article is an interpretation and sharing of the study →

Hyper.ai's official website updates this week:

* High-quality public datasets: 10
1️⃣ FewJoint small sample benchmark dataset
2️⃣ Wikipedia dataset
…

* AI4S paper cases: 3
1️⃣ The accuracy of early diagnosis of Parkinson's disease has increased to 90.2%. Shenzhen Institute of Advanced Technology and Zhongshan First Hospital jointly proposed the GSP-GCNs model
2️⃣ The paper of the Institute of Semiconductors of the Chinese Academy of Sciences was published in the top journal of TNNLS again, contributing a new perspective to explore mathematical expressions
…

* Popular encyclopedia entries: 10
1️⃣ Representation learning
2️⃣ Long Short-Term Memory
…

For more resources, please see this article →

AI for Science News at a Glance:

1️⃣ The world's first company to apply biomolecular condensates to drug discovery announced the layoff of 18 jobs
2️⃣ Google DeepMind releases AlphaGeometry
3️⃣ Evaxion Biotech announces the use of AI to develop customized cancer vaccines
4️⃣ iFlytek Medical officially submitted its application to the Hong Kong Stock Exchange
5️⃣ "Chongqing University-Baidu Intelligent Cloud Joint Innovation Center" was established

👀 For more useful information, please see this article →

Solving mathematical expressions is a very important research topic in the field of machine learning, and symbolic regression is a method of finding precise mathematical expressions from data. However, symbolic regression research also has its own difficulties, which makes it very time-consuming to find precise expressions.
In response to this academic problem, researchers from the Institute of Semiconductors of the Chinese Academy of Sciences regarded the solution of expression structure as a classification problem and solved it through supervised learning. They proposed a symbolic network called DeepSymNet to represent symbolic expressions. This article is a sharing and interpretation of the research →

Academic fraud is rampant despite repeated bans. The emergence of generative AI has lowered the threshold for fraud to a certain extent. The ability of chatbots such as ChatGPT to "tell lies seriously and justify themselves" has provided convenience for these people.
Not long ago, the Supervision Department of the Ministry of Science and Technology of my country issued the "Guidelines for Responsible Research Conduct (2023)", which clarified the boundaries of the reasonable use of generative artificial intelligence in scientific research projects. In addition, many well-known journals at home and abroad have also successively stated that they will restrict the various ways in which contributors use AI in the process of writing papers. This article further shares and interprets the new regulations and measures →

Data show that by 2030, the total number of Parkinson's disease patients in my country is expected to reach 5 million, almost half of the global number of patients. However, the pathogenesis of Parkinson's disease is still unknown, and only 20% cases can be attributed to specific genetic factors, so early diagnosis faces major challenges.
In response to this, researchers from the Shenzhen Institutes of Advanced Technology of the Chinese Academy of Sciences and the First Affiliated Hospital of Sun Yat-sen University proposed a graph signal processing-graph convolutional network (GSP-GCNs) model using the abnormal EEG characteristics of speech motor tasks in Parkinson's disease, which achieved high-precision intelligent diagnosis of Parkinson's disease. This article is an interpretation and sharing of the research results →

In order to make AI more efficient in music training, CCMusic has opened up some music and audio datasets for free use by computational musicology researchers, which are now available on hyper.ai. In addition, hyper.ai has also updated relevant music datasets from miHoYo, NetEase Cloud Music, etc. Let's take a look!

In recent years, the high threshold of the semiconductor industry has become increasingly prominent. In addition to the production process, which Huawei has already "demonstrated", chip design is also extremely challenging. Advanced chips such as the NVIDIA H100 Tensor Core GPU are composed of tens of billions of transistors and require multiple engineering teams to collaborate for two years to complete.
Interestingly, as an industry leader, Nvidia has also begun to think about using AI to make money more "easily". Not long ago, it released a customized large language model, ChipNeMo, which was trained based on its own internal data and can help engineers complete tasks related to chip design. This article is a detailed introduction to ChipNeMo →

In response to the high energy consumption of buildings, researchers proposed a new dynamic engineered multimodal feature learning (DEMMFL) model for long-term accurate prediction of building cooling loads, thereby achieving energy conservation goals. The model was applied to the building energy consumption of two office buildings in Hong Kong, and achieved good prediction accuracy and error performance. In the future, AI technology will be closely integrated with urban planning and management to create an efficient, sustainable and inclusive urban future.

HyperAI's new column is here~ Every Monday, the HyperAI editorial department will select the content (data sets, AI4S paper cases, encyclopedia entries) updated on the hyper.ai official website last week and publish them here. Welcome to visit hyper.ai directly to view all the content!

In recent years, with the increasing popularity of clean energy, solid-state batteries have received more widespread attention. Among them, proton conductive solid oxide batteries (P-SOC) have the advantages of low temperature operation and low ion conduction activation energy, and are gradually becoming known to people. However, a major obstacle to the development of high-performance P-SOC is the lack of efficient proton conductive air electrodes.
In response to this, researchers from Guangzhou University established a machine learning model based on the extreme gradient boosting (XGBoost) algorithm, which can be used for the screening of P-SOC air electrodes, and successfully screened out a high-efficiency air electrode material LCN91, whose activation energy is comparable to that of well-known air electrodes. This article is an interpretation and sharing of the experimental process →

Recently, Google DeepMind's Alpha series has added a new member - AlphaGeometry, which is still very popular, with praises such as "milestone", "epic", and "close to humans" overflowing the screen. So, how much gold is this AI system that claims to have gold medal-level mathematical ability?

The research team of Zhejiang University and Zhijiang Laboratory proposed a 3D molecular generation model based on protein pockets - ResGen, which can quickly generate drug-like molecules with lower binding energy and higher diversity. Compared with previous technologies, ResGen is 8 times faster and has been successfully applied to multiple links in drug design. The model adopts a parallel multi-scale modeling strategy to better consider the geometry of protein pockets and achieve higher computational efficiency. ResGen outperforms the existing most advanced models in multiple evaluation indicators, including binding energy, drug-like properties, etc. In real drug design scenarios, the molecules generated by ResGen have similar binding affinities to experimentally active molecules.

Luo Xiaozhou's team from the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, proposed a framework for predicting enzyme kinetic parameters (UniKP) to achieve the prediction of a variety of different enzyme kinetic parameters.

New achievements, new trends, and new perspectives of AI for Science - * AI pharmaceutical company spun off from DeepMind reaches first pharmaceutical cooperation worth $3 billion * Microsoft helps researchers discover 32 million new battery materials * TikTok is rumored to be in the United States […]

How does OpenAI's new "cash cow" make money? The launch of GPT Store has received great attention from users and the media. The media want to explore the deep meaning behind OpenAI's move, and some even directly compare it to Apple's App Store. Perhaps, simply comparing the App Store to guess OpenAI's intention is not accurate and comprehensive enough, so the HyperAI super neural editorial team started a bold brainstorming, hoping to catch clues in the GPT Store, which is "like fog, rain and wind".

Cities not only carry people's eager expectations for a peaceful and happy life, but are also an important foundation for supporting various economic activities. From the agricultural age to the industrial age, and then to today's digital age, people have never stopped improving the comfort and safety of cities. In this process, the importance of urban planning has become increasingly prominent.
The research team of Tsinghua University proposed a reinforcement learning model and method for urban community spatial planning, and realized an urban planning process in which human planners collaborate with artificial intelligence algorithms, providing a new idea for the automated planning of smart cities.

In 1928, the discovery of penicillin gave humans a powerful weapon to defeat pathogenic bacteria for the first time. However, the widespread use of antibiotics has also brought about a huge crisis - antibiotic resistance. The misuse of antibiotics has led to the emergence of certain "super bacteria", which has become an important clinical cause of disease in the 21st century. To solve this problem, it is urgent to develop new antibiotics.

Researchers from MIT developed a deep learning method to discover antibiotics, using the graph neural network Chemprop to identify potential antibiotics from a large chemical library and discovered a new class of antibiotics. This article is an interpretation and sharing of the experimental process →