Wang, J. et al. (2024). Mobile-Agent-v2: Mobile Device Operation Assistant with Effective Navigation via Multi-Agent Collaboration. arXiv preprint arXiv:2406.01014.
Wang, J., Xu, H., Jia, H., Zhang, X., Yan, M., Shen, W., … & Sang, J. (2024). Mobile-Agent-v2: Mobile Device Operation Assistant with Effective Navigation via Multi-Agent Collaboration. arXiv preprint arXiv:2406.01014.
X-PLUG. (2024). MobileAgent: The Powerful Mobile Device Operation Assistant Family. GitHub repository. https://github.com/X-PLUG/MobileAgent
ModelScope. (2024). ModelScope-Agent: An agent framework connecting models in ModelScope with the world. GitHub repository. https://github.com/modelscope/modelscope-agent
ALS 的基本思想是每次学习 $q$ 和 $p$ 中的一个进行优化,而将另一个保持不变。这使得每次迭代的目标函数都是凸的且可解的。当交替迭代收敛到最优解时,过程停止。值得注意的是,这种迭代计算可以并行化和/或分布式处理,这使得该算法在数据集很大且用户-物品评分矩阵非常稀疏的推荐场景中非常理想。关于 ALS 及其分布式计算的全面讨论,请参见这里。
import warnings
warnings.simplefilter(action='ignore', category=FutureWarning)
import sys
from matplotlib import pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns
import pyspark
from pyspark.sql import SparkSession
from pyspark.ml.recommendation import ALS
import pyspark.sql.functions as F
from pyspark.sql.functions import col
from pyspark.ml.tuning import CrossValidator
from pyspark.sql.types import StructType, StructField
from pyspark.sql.types import FloatType, IntegerType, LongType
from recommenders.datasets import movielens
from recommenders.utils.spark_utils import start_or_get_spark
from recommenders.evaluation.spark_evaluation import SparkRankingEvaluation, SparkRatingEvaluation
from recommenders.tuning.parameter_sweep import generate_param_grid
from recommenders.datasets.spark_splitters import spark_random_split
print(f"System version: {sys.version}")
print(f"Pandas version: {pd.__version__}")
print(f"PySpark version: {pyspark.__version__}")
为了解决上述问题,研究人员提出了一种新的对比学习框架CLSR(Contrastive Learning for Short and Long-term interests in Recommendation)。该框架的核心思想是:通过自监督的方式,利用用户的交互序列构建长短期兴趣的代理表示,并通过对比学习实现长短期兴趣的有效分离。具体来说,CLSR包含以下几个关键组件: