social practice> Types of section> Science and Technology Research Institute remote project: natural language processing direction
                            Science and Technology Research Institute remote project: natural language processing direction
                            Project Category: Science and Engineering Internship Research
                            Enrollment quota: 20 people
                            Suitable for groups: excellent undergraduates and sophomores with high computer skills
                                                        Activity time: annual enrollment
register now
(Is there still a question before registration? Welcome to call customer service: 010-57952000)

project's venue:

                                Beijing

Project purpose:

                                The project aims to enable students to learn the basic knowledge of natural language processing through a combination of theory and practice, to deeply study the relevant theoretical foundations, and to quickly grasp the deep mainstream platform.

project cost:

点击咨询>>

I. Project introduction

Natural language processing focuses on how to make machines better understand humans. Natural language processing technology has been widely used in the fields of automatic question and answer, machine translation and so on. Baidu secret, Microsoft Xiaobing Apple Baidu secret, Microsoft Xiaobing, Apple siri are the crystallization of natural language processing technology. In recent years, with the development of deep learning technology, natural language processing has attracted much attention in the field of artificial intelligence. How to use the deep learning model to better solve the core tasks in the field of natural language processing is the current research hotspot.

This course aims to comprehensively introduce the current research and application status of natural language processing, elaborate the design ideas of mainstream deep learning models, introduce the advantages of word vector representation, and combine theory with practice to guide students with text classification as an application example. Quickly master the depth learning platform and methods. After the end, the instructor will issue a letter of recommendation based on the student's performance.

Second, the project content

The project aims to enable students to learn the basic knowledge of natural language processing through a combination of theory and practice, to deeply study the relevant theoretical foundations, and to quickly grasp the deep mainstream platform. The course will comprehensively introduce the core issues and technical challenges of natural language processing, explain the deep learning, the current application status and development trend in text processing, and the mainstream deep learning model (taking convolutional neural networks as an example) from theoretical principles to specific Implement a deep analysis. In addition, the course will be classified into texts as application examples, guiding students to use deep learning models to solve practical problems, fostering innovative thinking and independent abilities, and laying a good foundation for the further application and application of natural language processing and deep learning.

Third, the teacher background

The instructor is an associate researcher at a well-known institution of the Chinese Academy of Sciences. The main direction of data mining, deep learning, social computing, the AAAI, ICDM, PAKDD and other well-known national conferences and journals

Published more than 30 papers. Serving as a PC conference and reviewer for several international conferences.

Fourth, enrollment objects and requirements

Senior undergraduates and high school students with higher computer skills, plan to apply for automation, computer, software engineering, interdisciplinary (such as business finance and other data analysis) related majors. In order to allow students to better complete research projects, the project team will screen students in the form of written tests and interviews.

V. Itinerary

The distance of the scientific research project is one month, and the specific time can be arranged according to the needs and progress of the students. The advantage of this project is that for students with long enough application time, the tutor can help students to complete one or more professional research tasks in a more in-depth, comprehensive and systematic manner, so that students can participate in the whole process of scientific research. Experience and solve the problem of accomplishment while giving students a deep understanding of the background and frontier dynamics of the field. In addition to regular research project discussion courses, students can ask relevant questions at any time during the period, and get the professional guidance of the tutor to let the students experience the real work and life status of a researcher in advance. The course schedule is as follows:

the first week Natural language processing basic knowledge learning: the concept, natural language processing basic knowledge learning: the concept, the basic knowledge of natural language processing learning: the concept, the core problems in the field of language processing, difficulties and challenges, but in the field of language processing Core issues, difficulties and challenges, the application status and development trends of natural language processing. Appointed time to communicate with students online, ask questions about Xi Zhongcun, and communicate with students online, and answer questions about Xizhong.
the second week Deep learning concepts, application and development platforms: basic, deep learning concepts, application and development platforms: basic, deep learning concepts, application and development platforms: basic, unified design framework for machine learning methods, several mainstream depth model ( The unified design framework of the convolutional machine learning method, several mainstream depth models (the unified design framework of the convolutional god machine learning method, the application fields and development of several mainstream depth models (convolution neural network, circulation god confrontation generation, etc.) Application fields of pre-network, cycle god confrontation generation, etc., and application fields such as pre-development network, cycle god confrontation generation, etc., and development fields of pre-development networks, cycle gods, etc.) The application field and development prospects of God's confrontation generation. Learn and master at least one depth development framework (such as tensorflowensorflowensorflow ensorflow platform). Appointment time and online communication with students, and answer questions about Xizhong.
The third week Word vector (vector word (Word Embeddingord Embedding ord Embeddingord Embeddingord Embeddingord Embeddingord Embeddingord Embeddingord Embedding) Explanation:) Explanation: The natural language processing task word representation model, comprising a bag of words, clustering and embedding methods described object representation model, includes the words Bag, clustering and embedding methods introduce the purpose representation model, including word bag, clustering and embedding method introduction purpose representation model, including word bag, clustering and embedding methods. Introduce several mainstream word embedding learning models (the first few mainstream the word design) embedded learning model (CBowBowBow, Skip -gramgram) design concept and main principles. Arrange students to learn relevant literature, agree on time and online communication and main principles. Arrange students to learn relevant literature, agree on time and online communication and main principles. Arrange students to learn relevant literature, agree on time and online communication and answer questions in the study.
the fourth week Text classification topic combat: the basic idea of ​​introduction, using word embedding to represent the text classification topic combat: the basic idea of ​​introduction, using word embedding to represent the text classification topic combat: the basic idea of ​​introduction, using word embedding representation and deep learning model to solve the problem The basic idea. The instructor implements a basic idea of ​​a and deep learning model to solve the problem. The instructor implements a text classification model for convolutional neural networks. Arrange the students to learn related articles, and the text classification model when agreeing. Arrange the students to learn related articles, and the text classification model when agreeing. Arrange for students to learn related offerings, communicate with students online at the appointed time, and answer questions about Xizhong.