Bui Tan Truong once wanted to be a “streamer” after his contract with Becaemx Binh Duong expired. (photo by Huu Pham) Not afraid of pressure, but worried about family influence “My life as a player has a lot to say,”-Tan Truong shared at the beginning of the livestream. The goalkeeper of Hanoi club said that he is a person who likes to exchange and talk. Tan Truong also talked about mistakes when doing the job: “Human, no one is perfect. If I’m too perfect, I’ll be arrested abroad. I’m human too, are you sure you’re not wrong all your life? If you comment, I will listen and correct it, but say I’m negative, I don’t care. After Dang Van Lam was not summoned to the Vietnamese team by Coach Park Hang-seo due to the impact of the COVID-19 epidemic, Tan Truong was selected to start 3 matches in the 2022 World Cup 2nd Qualifier in the UAE. The goalkeeper of Hanoi club played very well. Bui Tan Truong became the number 1 goalkeeper in the second qualifying round of the 2022 World Cup in the UAE after Dang Van Lam was not called up due to the COVID-19 epidemic. (photo by Huu Pham) He said he was very worried, because in the past he was criticized for making mistakes. “The matches are really nervous, because in the past, I often made mistakes, so I was afraid of being criticized by people. In the old days, I used to read the press, wrong about everyone with the criticizing newspaper. I can overcome it, but I’m afraid of psychology because of my children and family. As long as I play, I can overcome the pressure. But I am afraid of the feeling that my children hear others say that their father has sold out, being criticized. I am very afraid that my family will be affected, but I am fine,” said Bui Tan Truong. According to Bui Tan Truong, before joining the team and going to Hanoi, he played at the end of 2019, his contract with Becamex Binh Duong expired. He went home to rest for 7 months, thinking he wouldn’t play anymore. While playing for Becamex Binh Duong Club, Tan Truong made many mistakes and was criticized by fans. However, he is highly regarded for his expertise. (photo by Huu Pham) “I think I will do a livestream, each time it gets a hundred viewers. There was a time “live” for three or four hours that dozens of people watched. I appreciate them very much. There is a wave of hundreds, hundreds of people watching it, it’s really fun. Going to Hanoi has better interaction, but I’m less live. The highest hour is more than 12k viewers. Before, I actively interacted with viewers to get more viewers, but now many people read. Sometimes I like to do Vlog a lot, guys, but I can’t do it alone, I don’t know how. You must have an e-kip as well,” Tan Truong revealed. “9 years old selling lottery tickets, my life can be written into a book” Tan Truong shared that his childhood dream was to drive a car: “I am fortunate to be a professional footballer, earning money to support my family. When I was 9-10 years old, my dream was to be a car driver and earn money to support my family. Dreaming of being a driver, not a player.” According to Bui Tan Truong, he tries to work for his family, then if he is rich, he will contribute to the society, because “doing a lot of money can’t do anything, dying can’t take it away. It’s a lot to eat four meals a day.” And here is a part of Tan Truong’s autobiography about his life: “I grew up in a rural area, Long Hau commune, Dong Thap province. I grew up in Long Hau market. When I was about 9-10 years old, I used to sell lottery tickets, around 96 or 97 or something, lottery tickets were about 1,000/sheet. My family is not too poor, my mother is a trader from the past, living near the market. I sell lottery tickets because in my hometown from childhood to adulthood everyone knew gambling: playing cards, cockfighting, soccer… I grew up there and got infected. Tan Truong played well in 3 matches of the Vietnamese team in the UAE, especially in the match against Malaysia on 11/6. (photo by Huu Pham) When I was nine or ten years old, I knew how to gamble, the game for money is all about playing. Growing up, I knew that my family was difficult, since I was young I did not live with my father. Since childhood, she lived with her mother, raising 4 siblings. My mother did a lot of work, so from the moment I realized it, I loved her very much. Wherever my mother went to do business, I followed, when I was a child, I followed her mother. The furthest I went to Sadec. When I was twelve years old, I knew how to drive a Honda and smoke. When I was young, I wanted to try it, and when I saw adults smoking, I wanted to try it. At first I didn’t know I was still choking, then I became an addict.” Bui Tan Truong also shared many other memories when he was a child, such as playing in the sand, earning 12 thousand at first. Twelve years old driving 30-50kg of pork at 2am to help the family… The “livestream” of Tan Truong’s narrative until last night attracted more than 67,000 interactions and thousands of comments.

Illustration According to the RT , a link to a pornographic website was recently discovered in a textbook used to teach Ukrainian to 10th graders since 2018. Ukrainian children start school at the age of six. By the 10th grade, most of the students were 16 years old. “It’s not clear if this is a joke or a hoax. But can you imagine how many children we have raised in the last three years with such a book?”, said Prosecutor General Irina Venediktova angrily, adding that prosecutors are conducting the investigation. investigate the case. According to local media, the person who discovered the error in the textbook was the author Aleksandr Avramenko. This famous education expert took to Facebook to complain about the slowness of the Ukrainian authorities. He said the cybersecurity agency was slow to respond when he called for the blocking of pornographic websites linked to in textbooks. Regarding his error, Avramenko explained that when writing the book, he cited some documents from the website that was once the official portal of a district in the Kirovograd Region of Ukraine. Because of quoting information, Mr. Avramenko must specify the link according to copyright law. When the textbook was published in 2018, the site still belonged to the Novomyrhorod district. But last year, the website domain was transferred to a new owner. The person then started turning the site into a dark web, Avramenko stressed, adding that “this metamorphosis was just discovered.” The author suggested that the dark web scandal could have been a provocation by some unknown opponents who were jealous of his success. “Why is this happening to my book? This year, I submitted four textbooks to different competitions and all of them won first place, far ahead of the competitors. Could this be the reason? Unfortunately, I also have enemies,” Avramenko wrote. However, the author does not rule out the possibility that this is simply a coincidence. Currently, the Ukrainian investigative agency has not confirmed or denied the author’s allegations.

Sharing about the reason for choosing to travel independently, Ty said: “Perhaps it is because I want my youth to be a little different. Before every trip, I have a feeling of fear, but after overcoming it, I want to go to more places. And another reason, there is a sister I know who inspired me with self-sufficient travel trips. She’s a girl, but if she’s brave like that, I can do it too!” Self-sufficient travel helps Ty balance in his studies. (Photo: NVCC) Nguyen Ty affirmed that she had conquered her limits or the things she thought she couldn’t do, and then she wanted to conquer more and more goals. Talking about the trip that left the most emotions in Ty, the friend was not afraid to share about his trip to Y Ty (Lao Cai). “Initially, I went “trekking” (in the form of hiking or climbing mountains, crossing all kinds of terrain) Lao Than in Y Ty. Due to the influence of the storm, it rained and thanks to this chance, I met two very interesting people, Uncle Lee and Ms. Van in Y Ty. After sharing a meal, I invited them to join for fun. During this trip, I experienced a lot of things like the landscape of the high mountains, how life in the mountains lacks electricity because I only use electricity while eating the rest, I have to light a fire in the kitchen to keep warm. … And the trekking journey of Lao Than in the rainy and stormy season is not easy, I slipped and fell several times when climbing the slippery slopes. However, this trip left me with a hobby of trekking, maybe in the future I will continue to go to the mountains higher than Lao Than”, Ty recalled. Photo Bill “Chill” at a coffee shop in Da Lat. In such self-sufficient tours, Ty made many mistakes because of his lack of experience. Billion shared: “Once, I booked a room on the wrong date, but fortunately, I was supported to change the date at no extra cost. There were times when I didn’t preview the weather forecast, so when I got there, I didn’t get much harvest. And even the times of not carefully calculating in terms of health to visit many places.” Ty’s memorable climb to Lao Than and Y Ty with new friends. Also from those mistakes, Ty has learned many valuable lessons for himself. From scheduling and organizing a reasonable plan, managing expenses, expanding relationships and connecting with people, he also has the opportunity to learn more about the culture in many different lands. listen directly to the story told by the natives. “Go a day, learn a smart sieve” and so, Nguyen Ty becomes more and more active and confident to conquer new destinations in the future. “The most important thing is to determine the route. , calculate and reserve costs if risks arise. If it’s your first time experiencing, you can choose places that are a little closer to you to be less surprised, and slowly go further. Ask people who have traveled alone many times what questions you still have. In my opinion, in today’s open world, it’s not too difficult to look up and prepare for the schedule. Every time I prepare to go, I go to pages like: Skyscanner, Booking.com, Vexere, DSVN… to see flights, train tickets, bus tickets, accommodation …”, Bill “reveals” the experience for those who travel independently for the first time. Trieu Dao Nguyen Ty always knows how to balance traveling and studying, so during high school, you have achieved some “terrible” achievements such as: Traditional Olympic Silver Medal 30/4 times XXIV Geography, Traditional Olympic Gold Medal 30/4 times XXV in Geography, First prize for Excellent Student in the City in Biology.

Boeing suspends 737 MAX deliveries due to electrical problems. (Source: Rappler) The slow delivery process will affect Boeing’s cash flow, as airlines and customers often pay most of the order value after receiving the goods. The 737 MAX is Boeing’s best-selling plane, but has been grounded for 20 months after two crashes that killed 346 people. Deliveries of the aircraft resumed in November last year, after Boeing updated its flight control system. But now, about 100 737 MAX planes are still “on hold” because of an electrical fault, and it took Boeing longer than anticipated to fix the problem. Ed Pierson, a former Boeing production manager, said the electrical fault should have been discovered during the evaluation of the 737 MAX aircraft after the two crashes. He once “complained” to the US Federal Aviation Administration (FAA) for focusing only on reviewing the flight control system without checking for other problems in the production process. Boeing said it received an order for 25 planes last month, but 17 of the 737 MAX planes were cancelled, leaving the number of planes ordered down to just eight. (according to Rappler)

The land data source with a ‘age’ of nearly 20 years and a series of errors compared to the actual land use situation makes the compensation workers at the Long Thanh Airport project always face the risk. engine ‘do it wrong’. Officer working in compensation at Long Thanh Airport project measured to determine the level of compensation and support for a house of a household in Binh Son commune. Photo: Pham Tung * “Thrilled” with each profile According to the land use right certificate (now collectively referred to as the pink book) issued to Mr. Tran Van Huu (living in Phuoc Thien commune, H. Nhon Trach) in 1999 by the People’s Committee of H. Long Thanh, in 1999, Mr. Tran Van Huu is a person. have use rights to 15 land plots with a total area of ​​more than 2.1 ha in Binh Son commune, Long Thanh district. However, in 2019, when measuring and updating current land use to implement the policy of compensation and support under the project of Land Acquisition, Compensation, Support and Resettlement at Long Thanh Airport, Counting Team No. 1, belonging to the Compensation Team of Long Thanh Airport project, discovered that 2 of the 15 land plots that have been issued pink books are not under the use right of Mr. Tran Van Huu. This is just one of many cases where the error between the pink book has been issued to the people and the actual land use situation in the project area of ​​Long Thanh Airport. Mr. Le Van Tiep, Acting Chairman of People’s Committee of Long Thanh District, said that if in normal projects, the work of compensation and site clearance is only about 14 steps, in the project of Long Thanh Airport, due to the If the land is old, it must take 20 steps to ensure accuracy and coherence. Mr. Bui The Su, a second officer of the Department of Natural Resources and Environment, to work at the Counting Team No. 1, under the Project Compensation Team of Long Thanh Airport, said that the Long Thanh Airport project had a very strong planning policy. After a long “waiting” period, data on land in the project area is not updated or supplemented, making errors in the origin and current land use occur very common. “From 2003 onwards, pink books issued to people were often written by hand instead of printed today. At the same time, the technology system for land data management was also outdated, so the overlapping and wrong allocation of land plots happened a lot ”- Mr. Bui The Su said. Seconded to support the implementation of land acquisition and site clearance for the Long Thanh Airport project from March 2019, over a year of working, Mr. Bui The Su said that there were errors in land data. Belt still occurs frequently in projects, however, large numbers such as in the Long Thanh Airport project are very few. “The reason, apart from outdated land data, is also due to the area of ​​land to be acquired, the number of affected households in the Long Thanh Airport project is also very large”. Many errors in land data put pressure on the compensation workers at the Long Thanh Airport project. In particular, the heaviest pressure is the risk of incorrectly confirming the current land use status, leading to the implementation of the policy of compensation and support for the wrong subjects. “Although I have done the verification and verification very carefully, each time I put a pen to sign to confirm the record, I still feel nervous” – Mr. Bui The Su shared. Both ensuring the progress of the compensation work and avoiding errors in the verification and confirmation of the file is the biggest challenge that the team of cadres and civil servants doing the compensation work at the Long Thanh Airport project. must face to face. Mr. Le Nguyen Hoang An, Head of Compensation Team of Long Thanh Airport project, said that for dossiers with a discrepancy between the pink book and current land use, the settlement usually takes a lot of time. Accordingly, the enumeration teams must coordinate with the Commune People’s Committee to accurately verify the current land use status. After that, the parties will invite the person whose name is on the pink book and the actual land user to work and negotiate the current land use status. When completed, the authorities must withdraw the issued pink book and decide to notify the land acquisition previously issued to issue a new decision on land acquisition in accordance with the current status. “In each case as above, the implementation process takes from 8 months to 1 year. Meanwhile, the pressure on site clearance progress to hand over land to investors is huge. Therefore, the compensation team has to work almost non-stop for more than 2 years ”- Mr. Le Nguyen Hoang An said. * Mobilizing people to consensus One of the biggest challenges facing the compensation workers at the Long Thanh Airport project is the discrepancy in land area on the pink book of the people and the actual situation. Mr. Bui The Su (left) and Mr. Le Nguyen Hoai An exchange to process a claim file of Long Thanh Airport project. Photo: Pham Tung The main reason for the difference in land area also comes from the “old” land data system in the project area of ​​Long Thanh Airport. The project area of ​​Long Thanh Airport has been planned for a long time, so the cadastral map is still a map that was made nearly 20 years ago. Specifically, the cadastral map in the planning area of ​​Long Thanh airport was made in 1996 according to the old aerial image measurement technology, the accuracy is not high. Meanwhile, according to the 2013 Land Law, the implementation of the policy of compensation and support when the State recovers land must be based on the actual area being used. Therefore, to carry out the compensation work, the entire land area of ​​the Long Thanh Airport project must be re-measured according to the current new technologies to ensure accuracy. Through the measurement process, the misalignment of the land area in the project area occurs a lot. “The number of dossiers for the difference in the area issued on the pink book compared to the current land use situation is up to hundreds of cases” – said Mr. Le Nguyen Hoai An. According to Mr. Le Nguyen Hoai An, if the land area on the pink book is equal to or larger than the actual land area, it is easy, but if it is smaller, it is extremely complicated. Because, people always base on the issued pink book to determine the area of ​​land under their use rights. Explaining and mobilizing people to accept “losing land” to correct the current land use situation is very arduous. “An inch of land. Therefore, in addition to the implementation of the compensation work, the people doing the compensation work in the project of Long Thanh Airport also have to also work on the civil service ”- Mr. An shared. Taking the evidence for the “double” task that the compensation workers have to perform in the Long Thanh Airport project, Mr. Le Nguyen Hoai An said that up to now, the Compensation Team of Long Thanh Airport project has recognized the most likely case of “losing” land is that of a household in Hamlet 7, Binh Son Commune when the actual land area decreases by 2 thousand m2 compared to the area granted on the pink book. Therefore, it took nearly a year with dozens of meetings between the authorities and the people. Just presenting the actual evidence after measurement, combined with propaganda and advocacy, the above households agree to complete the dossier. “Accepting to reduce 2 thousand m2 of land is accepting to lose hundreds of millions of money in compensation and support, so it is not easy for people to agree” – Mr. Le Nguyen Hoai An shared. Dong Nai has mobilized a contingent of cadres, civil servants and public employees up to 142 people to participate in the work of compensation and site clearance for the Long Thanh Airport project. Among these, there are 86 cadres, civil servants and public employees, who are seconded from agencies, departments, branches of the province and districts in the province; 7 cadres, civil servants and officials from the land fund development centers of districts and 49 cadres, civil servants and employees of the Hong Thanh Land Fund Development Center. Pham Tung Lesson 2: Historical “immigration”

Heart of the Machine Editorial Department Yuncong Technology Speech Group proposed a semantic error correction technical solution based on the BART pre-trained model. It can not only correct common spelling errors in ASR data, but also correct common sense errors, grammatical errors, and even some that require reasoning. Errors are corrected. In recent years, with the development of automatic speech recognition (ASR) technology, the recognition accuracy has been greatly improved. However, there are still some errors that are very obvious to humans in the ASR transcribing results. We don’t need to listen to the audio, we can find out only by observing the transcribed text. The correction of such errors often requires some common sense and grammatical knowledge, and even the ability of reasoning. Thanks to the recent development of unsupervised pre-training language model technology, error correction models based on plain text features can effectively solve such problems. The semantic error correction system proposed in this paper is divided into two modules: encoder and decoder. The encoder focuses on understanding the semantics of the output text of the ASR system, and the design of the decoder focuses on the use of standardized vocabulary to re-express. Link to the paper: https://arxiv.org/abs/2104.05507 introduction Text error correction is an important method to improve the accuracy of ASR recognition. Common text error correction includes grammatical error correction and spelling error correction. Because the error distribution of the ASR system transfer is quite different from the above error distribution, these models are often not suitable for direct use in the ASR system.Here, Yuncong Technology Speech Group proposed a pre-training model based on BART [1] The Semantic Error Correction (SC) technical solution can not only correct common spelling errors in ASR data, but also correct some common sense errors, grammatical errors, and even some errors that require reasoning. In our experiments with 10,000 hours of data, the error correction model can relatively reduce the error rate (CER) based on 3gram decoding results by 21.7%, achieving an effect similar to RNN rescoring. Using error correction on the basis of RNN re-scoring, a further 6.1% relative reduction in CER can be achieved. The error analysis results show that the actual error correction effect is better than the CER indicator shows. model 1) ASR semantic error correction system design The ASR semantic error correction process is shown in Figure 1. The semantic error correction module can be directly applied to the first pass of the decoding result as an alternative to the re-scoring module. In addition, it can also be connected to the re-scoring model to further improve the recognition accuracy. Figure1 ASR system with integrated semantic error correction model 2) Baseline ASR system The baseline acoustic model structure selected by the author is pyramidal FSMN[2], Training on 10,000 hours of Mandarin audio data. The WFST used in the first pass of decoding is composed of 3gram language model, pronunciation dictionary, dual phoneme structure and HMM structure. 4grams and RNNs are used in the rescoring, and the training data is the reference text corresponding to these audios.Acoustic model and language model use Kaldi tools [3] training. 3) Semantic error correction model structure The semantic error correction model proposed by the researcher is based on Transformer [4]Structure, it contains 6 layers of encoder layer and 6 layers of decoder layer, and the modeling unit is token. In the training process using the Teacher forcing method, the text output by the ASR is input to the input side of the model, and the corresponding reference text is input to the output side of the model. The input embedding matrix and the output embedding matrix are used for encoding, and the cross entropy is used as the loss function. In the semantic error correction model inference process, beam search is used for decoding, and the beam width is set to 3. Figure 2 Semantic error correction model based on Transformer experiment 1. Error correction training data preparation The training set of our baseline ASR model is 10,000 hours of Mandarin speech data, which contains about 800 transliterated texts. The side test set consists of 5 hours of mixed voice data, including Aishell, Thchs30 and other side test sets. In order to fully sample the error distribution identified by the ASR system, we adopted the following techniques when constructing the error correction model training data set: Use a weak acoustic model to generate error correction training data. Here, 10% of the speech data is used to train a small acoustic model separately to generate training data; Add disturbance to the MFCC feature, and randomly multiply the MFCC feature by a coefficient between 0.8 and 1.2; Input the noisy features into the weak acoustic model, take the first 20 results of beam search, and filter the samples according to the typos rate threshold. Finally, we pair the filtered decoding results with their corresponding reference texts as the error correction model training data. By decoding the full audio data and setting the threshold at 0.3, we obtained about 30 million error correction sample pairs. 2. Input and output presentation layer In the semantic error correction model, the input and output text use the same dictionary. But the typo in the input text contains more semantics than its standard usage, while the output text only uses the standard words to express. Therefore, the independent representation of the tokens on the input and output sides is more in line with the needs of error correction tasks. The results in Table 1 prove our inference. The experimental results show that when the input and output embedded matrix share the weight, the error correction model will bring negative effects. When the input and output tokens are represented independently, the CER of the system can be reduced by 5.01%. 3. BART vs BERT initialization Here, the researcher pre-trains the language model technology, and transfers the semantic knowledge learned from the large-scale corpus to the error correction scene, so that the error correction model obtains better robustness and universality on a relatively small training set.化性. We compare random initialization, BERT[5]Initialization and BART[1]Initialization method. During the initialization process, because the BART pre-training task and model structure are the same as Transformer, the parameters can be reused directly.In BERT initialization, both the encoder and decoder of Transformer are applicable to the first 6-layer network parameters of BERT[6]. The results in Table 2 show that BART initialization can reduce the typo rate of the baseline ASR by 21.7%, but the improvement of the BERT-initialized model relative to the random initialization model is very limited. We push this may be because the structure of the BERT and the semantic error correction model and the training target are too different, and the knowledge has not been effectively transferred. In addition, the error correction model corrects the output of the language model after re-scoring, and the recognition rate can be further improved. Compared with 4grams, RNN re-scoring results, CER can be relatively reduced by 21.1% and 6.1%, respectively. 4. Error correction model vs large language model Generally speaking, the ASR system uses a larger language model to obtain better recognition results, but it also consumes more memory resources and reduces decoding efficiency. Here, we add a large number of crawlers or open source plain text corpus on the basis of the speech data reference text, and newly train 3gram, 4gram and RNN language models, and call them the big language model. The one used in the baseline ASR system is called a small model. By comparison, it is found that the recognition accuracy of adding error correction on the basis of a small model surpasses the effect of using a large model alone. In addition, by using semantic error correction on the basis of a large model, the recognition rate can be further improved. Some examples of error correction are as follows: 5. Error analysis In error analysis of 300 examples of failed corrections, we found that the actual effect of semantic error correction is significantly better than the CER indicator evaluation. About 40% of errors hardly affect semantics, for example, some transliterated foreign names or place names There are many ways of expression, some personal pronouns lack context, which will cause the mixed use of “her, other, and other”, and some are substitutions of modal particles that do not affect semantics. In addition, 30% of errors are not suitable for correction based on pure text features due to insufficient contextual information. The remaining 30% of errors are caused by insufficient semantic understanding or expressive ability of the semantic error correction model. to sum up This paper proposes a BART-based semantic error correction model, which has good generalization and consistently improves the recognition results of multiple ASR systems. In addition, the researchers verified the importance of independent representation of input and output in the task of text error correction through experiments. In order to more fully sample the ASR system identification error distribution, this paper proposes a simple and effective error correction data generation strategy. Finally, although the semantic error correction method we proposed has achieved certain benefits, there is still room for optimization, such as: 1. The introduction of acoustic features helps the model identify whether there are errors in the text and reduces the false touch rate. 2. Introducing more contextual information can eliminate some semantic ambiguities or missing information in the text. 3. Adapt to vertical business scenarios to improve the recognition accuracy of some professional terms. references [1]M. Lewis et al., “BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension,” 2019, [Online]. Available: http://arxiv.org/abs/1910.13461. [2]X. Yang, J. Li, and X. Zhou, “A novel pyramidal-FSMN architecture with lattice-free MMI for speech recognition,” vol. 1, Oct. 2018, [Online]. Available: http://arxiv.org/abs/1810.11352. [3]D. Povey et al., “The Kaldi Speech Recognition Toolkit,” IEEE Signal Process. Soc., vol. 35, no. 4, p. 140, 2011. [5]J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding,” 2018, [Online]. Available: http://arxiv.org/abs/1810.04805. [6]O. Hrinchuk, M. Popova, and B. Ginsburg, “Correction of Automatic Speech Recognition with Transformer Sequence-to-sequence Model,” ICASSP 2020-2020 IEEE Int. Conf. Acoust. Speech Signal Process., pp. 7074C7078, Oct . 2019, [Online]. Available: http://arxiv.org/abs/1910.10697. Build new, see wisdom – 2021 Amazon Cloud Technology AI Online conference April 22, 14:00-18:00 Why do so many machine learning loads choose Amazon Cloud Technology? How to achieve large-scale machine learning and enterprise digital transformation? “Building New · Seeing Wisdom-2021 Amazon Cloud Technology AI Online Conference” is led by Alex Smola, vice president of global artificial intelligence technology and outstanding scientist of Amazon Cloud Technology, and Gu Fan, general manager of Amazon Cloud Technology Greater China Product Department, and more than 40 heavyweights Guests will give you an in-depth analysis of the innovation culture of Amazon cloud technology in the keynote speech and 6 major conferences, and reveal how AI/ML can help companies accelerate innovation. Session 1: Amazon Machine Learning Practice Revealed Session 2: Artificial Intelligence Empowers Digital Transformation of Enterprises Session 3: The Way to Realize Large-scale Machine Learning Session 4: AI services help the Internet to innovate rapidly Session 5: Open Source and Frontier Trends Sub-venue 6: Intelligent ecology of win-win cooperation Which topic are you more interested in in the 6 major conference venues?