CN116978462A - Method for generating non-natural promoter based on diffusion model - Google Patents
Method for generating non-natural promoter based on diffusion model Download PDFInfo
- Publication number
- CN116978462A CN116978462A CN202310954854.2A CN202310954854A CN116978462A CN 116978462 A CN116978462 A CN 116978462A CN 202310954854 A CN202310954854 A CN 202310954854A CN 116978462 A CN116978462 A CN 116978462A
- Authority
- CN
- China
- Prior art keywords
- promoter
- generating
- diffusion model
- promoters
- region
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000009792 diffusion process Methods 0.000 title claims abstract description 35
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000013135 deep learning Methods 0.000 claims abstract description 6
- 238000012549 training Methods 0.000 claims description 25
- 108090000623 proteins and genes Proteins 0.000 claims description 18
- 108091026890 Coding region Proteins 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 8
- 108091092724 Noncoding DNA Proteins 0.000 claims description 6
- 238000013527 convolutional neural network Methods 0.000 claims description 6
- 230000007246 mechanism Effects 0.000 claims description 4
- 239000003814 drug Substances 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 239000013598 vector Substances 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 abstract description 2
- 230000009286 beneficial effect Effects 0.000 description 5
- 238000013461 design Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000035772 mutation Effects 0.000 description 2
- 238000013518 transcription Methods 0.000 description 2
- 230000035897 transcription Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 241000588724 Escherichia coli Species 0.000 description 1
- 241000206602 Eukaryota Species 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000003042 antagnostic effect Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 238000009510 drug design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012269 metabolic engineering Methods 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 230000014493 regulation of gene expression Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005549 size reduction Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B5/00—ICT specially adapted for modelling or simulations in systems biology, e.g. gene-regulatory networks, protein interaction networks or metabolic networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Biotechnology (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Spectroscopy & Molecular Physics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Public Health (AREA)
- Physiology (AREA)
- Bioethics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Epidemiology (AREA)
- Databases & Information Systems (AREA)
- Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
Abstract
The application discloses a method for generating a non-natural promoter based on a diffusion model, and belongs to the technical field of biological information. In order to realize the generation of the promoter, a deep learning network based on a diffusion model is established. Meanwhile, the application carries out true and false judgment and functional interval analysis on the generated promoters, and the results show that more than 40% of the generated promoters are true promoters, and the sequences have obvious-35 and-10 functional regions and have higher credibility.
Description
Technical Field
The application relates to a method for generating a non-natural promoter based on a diffusion model, and belongs to the technical field of biological information.
Background
The promoter design can assist in the construction of metabolic engineering networks for de novo synthesis of chemicals, pharmaceuticals and other raw materials in microorganisms. The promoter mainly plays a role in promoting gene transcription and translation and has visual influence on the expression quantity of a target gene. Recent researches show that the pearson correlation coefficient of the transcription amount and the translation amount of the promoter gene is up to 0.8, so that accurate regulation of protein expression can be realized by regulating the promoter. In previous studies, researchers have tried different methods for mining non-natural promoters, including directed evolution schemes, randomly introducing mutation sites into the target promoter, and rational design, i.e., mutation of only a small segment of the conserved or non-conserved regions of the promoter.
At present, although a certain progress is made in non-natural promoter screening, the constructed promoter library is still small. In general, the promoter is 50 bases long and has 450 composition modes, and is difficult to verify by only experimental methods. The length of the promoter of eukaryote is far more than 50 bases, and the difficulty in experimental screening is greater. Therefore, it is extremely important to develop a method for computationally assisting in promoter generation, which will facilitate the screening of promoters.
Wang et al in 2020 proposed to realize de novo design of promoters against a generation network, convert promoter genes into one-dimensional arrays for learning, and further generate a brand new artificial molecule sequence similar to natural biomolecules in distribution through self-game of a generator and a discriminator, thereby realizing de novo design of promoters. However, the countermeasure generation network has great instability in training due to the contradiction between the training optimal discriminant and the minimization generator, and the diversity of the promoters generated by countermeasure learning is limited to a certain extent, so that the countermeasure generation network is not easy to expand to modeling complex multi-modal distribution. For the above reasons, it is necessary to investigate a novel method for generating a non-natural promoter.
Disclosure of Invention
In order to solve the problem of instability in the de novo design of promoters currently implemented against the production network, the present application provides a method for generating non-native promoters based on a diffusion model, comprising:
step S1: constructing a diffusion model for generating the non-natural promoter, wherein the diffusion model for generating the non-natural promoter depends on UNet in a convolutional neural network, and the convolutional neural network is adopted when a coding region of the UNet is built; the non-coding region adopts an up-sampling mode to restore the image size; carrying out feature transfer by adopting a normative UNet jump connection between the coding region and the non-coding region, and introducing self-attention mechanisms into the coding region and the decoding region;
step S2: training the diffusion model for generating the non-natural promoter by adopting the promoter in the public data set as training data;
step S3: and generating a new promoter by using a trained diffusion model for generating the non-natural promoter.
Optionally, the step S2 includes:
digitizing the gene sequence of the promoter in the public dataset;
training the diffusion model for generating the non-natural promoter by using the gene sequence of the promoter after the digitization treatment, calculating a loss value in the training process, carrying out promoter identification and conservation assessment on an output sample, and storing model parameters after the training is completed;
the promoter identification adopts a deep learning-based Promo R module to judge the true and false of each generated sequence, and calculates the proportion of the true promoter to all the generated promoters;
the conservation of the promoter is evaluated by aligning the sequences of the generated promoter and observing the sequences of the-35 region and the-10 region, and when the identifiers of the-35 region and the-10 region are TT and TATAAT, the generated promoter is considered to have the characteristics of the natural promoter.
Alternatively, the method is performed using MetaLogo as a tool for observing the sequences of the-35 and-10 regions.
Optionally, the digitizing the gene sequence of the promoter in the public dataset comprises:
extracting features by adopting a single-heat coding method, and converting a sequence with the length of 50 bases into vectors with the number of channels of 1, the length of 4 and the width of 50;
the transformed bases A, T, C, G are respectively: [ 10 00], [ 00 01 ], [ 01 00], [ 00 10 ].
Optionally, the method further comprises setting a ratio threshold of true promoters to all promoters generated.
The application also provides application of the method for generating the non-natural promoter based on the diffusion model in chemicals and medicines.
The application has the beneficial effects that:
the digitization of the gene sequence is realized by computer coding, and the public data set is collected to construct a training data set. The application uses digitized gene as input and adopts diffusion model to learn its characteristic, and evaluates the quality of the generated sample and uses it to generate non-natural promoter. And has no defect of unstable training. Meanwhile, the diffusion model is more beneficial to stably generating promoters with higher diversity and is beneficial to excavating new promoters.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of generating promoters based on a diffusion model provided in one embodiment of the present application.
FIG. 2 is a sequence identification diagram of a promoter generated by using a diffusion model according to the present application.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present application more apparent, the embodiments of the present application will be described in further detail with reference to the accompanying drawings.
PyTorch: is the python version of torch, a neural network framework that is open-sourced by Facebook, specifically programmed for GPU-accelerated Deep Neural Networks (DNNs). Torch is a classical tensor (tensor, a digital container in a machine learning program, essentially an array of various dimensions, usually called axes, the number of axes being called orders) library operating on multidimensional matrix data, and has wide application in machine learning and other mathematics intensive applications. Unlike the static computational graph of Tensorflow, the computational graph of pytorch is dynamic and can be changed in real time according to the computational needs.
Embodiment one:
the present embodiment provides a method for generating a non-natural promoter based on a diffusion model, which performs diffusion model training based on a promoter gene sequence and generates a new promoter, see fig. 1, and the method includes:
step S1: normalization of the input gene sequence.
Building a training set: the promoter in the dataset reported by Thomason (which is the E.coli promoter) was used as training data. The training set contains 11884 promoters in total;
step S2: digitization of the gene sequence of the input promoter.
And (3) performing feature extraction by adopting a single-heat coding method, and converting a sequence with the length of 50 bases into vectors with the channel number of 1, the length of 4 and the width of 50. The transformed bases A, T, C, G are respectively: [ 10 00], [ 00 01 ], [ 01 00], [ 00 10 ].
By this step the gene sequence can be subjected to a digitization process, such as a promoter:
CCGCTCAAATATTGTTAAATTGCCGGTTTTGTATCAACTACTCACCCGGG is converted into: [[0 1 00][0 10 0][0 01 0][0 10 0][0 00 1][0 10 0][1 00 0][1 00 0][1 00 0][0 00 1][1 00 0][0 00 1][0 00 1][0 01 0][0 00 1][0 00 1][1 00 0][1 00 0][1 00 0][0 00 1][0 00 1][0 01 0][0 10 0][0 10 0][0 01 0][0 01 0][0 00 1][0 00 1][0 00 1][0 00 1][0 01 0][0 00 1][1 00 0][0 00 1][0 10 0][1 00 0][1 00 0][0 10 0][0 00 1][1 00 0][0 10 0][0 00 1][0 10 0][1 00 0][0 1 00][0 10 0][0 10 0][0 01 0][0 01 0][0 01 0]].
Step S3: the diffusion model is constructed for sequence feature learning, and specifically comprises the following steps:
s3.1: converting the data into a PyTorch identifiable tensor;
mainly converting Numpy array into tensor, which can be referred to
The transformation is performed in the transformation mode described in https:// blog.
S3.2: constructing a deep learning network based on PyTorch, wherein the network construction is mainly based on UNet in a convolutional neural network, and the convolutional neural network is adopted when a coding region of the UNet is constructed; the non-coding region performs image size reduction by means of upsampling. Carrying out feature transfer by adopting a normative UNet jump connection between the coding region and the non-coding region, and introducing self-attention mechanisms into the coding region and the decoding region;
diffusion model: under the condition that the diffusion process is a Markov chain, the Gaussian noise is continuously added to the original information, and the process of adding the Gaussian noise at each step is from X t-1 →X t ，
Wherein X is t Data representing time t-1, X t-1 Data at time t after the addition of gaussian noise is shown.
The back-diffusion process is to recover the original data from gaussian noise, assuming that the back-diffusion process is still a markov chain process, what is to be done is X T →X 0 Wherein X is T Data representing T time after adding Gaussian noise, X 0 Refers to the original data recovered from the data at time T after the addition of gaussian noise.
UNet: is a typical encoder, decoder architecture, with the encoder mainly performing feature extraction and the image size continually decreasing. And the right corresponds to the up-sampling process, by jumping the link with the information of the different convolution layers to restore the image to a size close to the original image.
Convolution layer: feature extraction is performed on the digitized promoters based on the image information. The formula is as follows:
an activation layer: the active layer uses a ReLU function, which can be understood as a piecewise linear function, changing all negative values to 0, while positive values are unchanged.
Self-attention layer: the self-attention mechanism allows inputs to interact with each other and find objects that they should focus more on, the output being the sum of these interactions and the attention score.
S3.3: and calculating a loss value in the model training process, and carrying out promoter identification and conservation assessment on the output sample.
The promoter identification adopts a Promo R module based on deep learning (BioRxiv: doi: https:// doi.org/10.1101/2023.03.05.531155), namely, each generated sequence is subjected to true and false discrimination, and the proportion of the true promoter to all generated promoters is calculated.
The promoter conservation was mainly performed on the generated promoter, and the sequences of the-35 and-10 regions were observed, using the tool Metalogo (http:// MetaLogo omicsnet. When the-35 and-10 regions are identified as TT and TATAAT, the resulting promoter is considered to be characteristic of the native promoter.
The loss value calculation adopts average square error loss (MSELoss) to compare the output result with the real result, and parameter optimization is realized through an AdamW optimizer.
S3.4: when the loss value in the training process is slowly reduced, evaluating a generated sample, judging according to a promoter identification result, and when a true promoter occupies a relatively high sequence with a significant conservation interval, storing training parameters of the training generation;
step S4: and generating the non-natural promoter by adopting a diffusion model according to the parameters stored after training.
In order to illustrate the superiority of the generated model constructed by the application, the application adopts the discrimination of true and false promoters based on deep learning and the characteristic analysis of the sequence of the generated promoter. The results are shown in FIG. 2, and it can be seen from FIG. 2 that the true duty ratio of the generated promoter is more than 40%, and the promoter has significant-35 and-10 regions, which indicates that the generated promoter has higher reliability.
According to the method for generating the natural thermal promoter based on the diffusion model, the digitization is realized by carrying out computer coding on the gene sequence, the digitization gene is used as input, the diffusion model is adopted for learning the characteristics, the quality of a generated sample is evaluated and used for generating the non-natural promoter, and compared with the existing technology for designing the promoter by antagonizing the generation network, the method for generating the natural thermal promoter based on the diffusion model is more stable in training, can effectively identify small-section key regions of the sequence, and can simultaneously identify the association of the small regions and the full-length sequence. And has no defect of unstable training. Meanwhile, the diffusion model is more beneficial to stably generating promoters with higher diversity and is beneficial to excavating new promoters.
Some steps in the embodiments of the present application may be implemented by using software, and the corresponding software program may be stored in a readable storage medium, such as an optical disc or a hard disk.
The foregoing description of the preferred embodiments of the application is not intended to limit the application to the precise form disclosed, and any such modifications, equivalents, and alternatives falling within the spirit and scope of the application are intended to be included within the scope of the application.
Claims (6)
1. A method of generating a non-native promoter based on a diffusion model, the method comprising:
step S1: constructing a diffusion model for generating the non-natural promoter, wherein the diffusion model for generating the non-natural promoter depends on UNet in a convolutional neural network, and the convolutional neural network is adopted when a coding region of the UNet is built; the non-coding region adopts an up-sampling mode to restore the image size; carrying out feature transfer by adopting a normative UNet jump connection between the coding region and the non-coding region, and introducing self-attention mechanisms into the coding region and the decoding region;
step S2: training the diffusion model for generating the non-natural promoter by adopting the promoter in the public data set as training data;
step S3: and generating a new promoter by using a trained diffusion model for generating the non-natural promoter.
2. The method according to claim 1, wherein the step S2 comprises:
digitizing the gene sequence of the promoter in the public dataset;
training the diffusion model for generating the non-natural promoter by using the gene sequence of the promoter after the digitization treatment, calculating a loss value in the training process, carrying out promoter identification and conservation assessment on an output sample, and storing model parameters after the training is completed;
the promoter identification adopts a deep learning-based Promo R module to judge the true and false of each generated sequence, and calculates the proportion of the true promoter to all the generated promoters;
the conservation of the promoter is evaluated by aligning the sequences of the generated promoter and observing the sequences of the-35 region and the-10 region, and when the identifiers of the-35 region and the-10 region are TT and TATAAT, the generated promoter is considered to have the characteristics of the natural promoter.
3. The method of claim 2, wherein the method is performed using a Metalogo as the means for observing the sequences of regions-35 and-10.
4. The method of claim 3, wherein digitizing the gene sequence of the promoter in the public dataset comprises:
extracting features by adopting a single-heat coding method, and converting a sequence with the length of 50 bases into vectors with the number of channels of 1, the length of 4 and the width of 50;
the transformed bases A, T, C, G are respectively: [ 10 00], [ 00 01 ], [ 01 00], [ 00 10 ].
5. The method of claim 4, further comprising setting a threshold ratio of true promoters to all promoters produced.
6. Use of the method for generating non-native promoters based on diffusion models according to any one of claims 1-5 in chemicals, pharmaceuticals.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310954854.2A CN116978462A (en) | 2023-07-31 | 2023-07-31 | Method for generating non-natural promoter based on diffusion model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310954854.2A CN116978462A (en) | 2023-07-31 | 2023-07-31 | Method for generating non-natural promoter based on diffusion model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116978462A true CN116978462A (en) | 2023-10-31 |
Family
ID=88476400
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310954854.2A Pending CN116978462A (en) | 2023-07-31 | 2023-07-31 | Method for generating non-natural promoter based on diffusion model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116978462A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118038993A (en) * | 2024-04-11 | 2024-05-14 | 云南师范大学 | Protein sequence diffusion generation method based on generation countermeasure network drive |
-
2023
- 2023-07-31 CN CN202310954854.2A patent/CN116978462A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118038993A (en) * | 2024-04-11 | 2024-05-14 | 云南师范大学 | Protein sequence diffusion generation method based on generation countermeasure network drive |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113971209B (en) | Non-supervision cross-modal retrieval method based on attention mechanism enhancement | |
| Liu et al. | Optimization-based key frame extraction for motion capture animation | |
| Korfmann et al. | Deep learning in population genetics | |
| CN116978462A (en) | Method for generating non-natural promoter based on diffusion model | |
| CN114023376B (en) | RNA-protein binding site prediction method and system based on self-attention mechanism | |
| CN103246922A (en) | Method for video abstract generation | |
| CN117153294B (en) | Molecular generation method of single system | |
| CN116312748A (en) | Enhancer-promoter interaction prediction model construction method based on multi-head attention mechanism | |
| CN116305939A (en) | High-precision inversion method and system for carbon water flux of land ecological system and electronic equipment | |
| CN118038032A (en) | Point cloud semantic segmentation model based on super point embedding and clustering and training method thereof | |
| EP4290524A2 (en) | Artificial intelligence-based epigenetics | |
| Bustillo | Enhanced Genetic Algorithm for Variable Minimization through Modified Crossover Operator | |
| Olariu et al. | Biology-derived algorithms in engineering optimization | |
| Zhao et al. | A novel hybrid GA/SVM system for protein sequences classification | |
| CN105069323A (en) | Memetic algorithm based microbial fermentation control and optimization method | |
| CN115169366A (en) | Session recommendation method based on sampling convolution and interaction strategy | |
| CN113343589B (en) | Genetic-random constant-based acidic natural gas hydrate generation condition prediction method adopting genetic expression programming | |
| CN113377907B (en) | End-to-end task type dialogue system based on memory mask self-attention network | |
| CN115660038A (en) | Multi-stage integrated short-term load prediction based on error factors and improved MOEA/D-SAS | |
| CN115545038A (en) | Aspect emotion analysis method for optimizing grid label | |
| CN115374854A (en) | Multi-modal emotion recognition method and device and computer readable storage medium | |
| CN112906527B (en) | Finger vein biological key generation method based on deep neural network coding | |
| CN109255020B (en) | Method for solving dialogue generation task by using convolution dialogue generation model | |
| CN113011519A (en) | Multi-scale classified data mining method | |
| Li et al. | Generative Design of Crystal Structures by Point Cloud Representations and Diffusion Model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |