AAAI2023「Are Transformers Effective for Time Series Forecasting?」と、HuggingFace「Yes, Transformers are Effective for Time Series Forecasting (+ Autoformer)」の紹介です。
AAAI2023「Are Transformers Effective for Time Series Forecasting?」と、HuggingFace「Yes, Transformers are Effective for Time Series Forecasting (+ Autoformer)」の紹介です。
表形式データのために提案されたDNNをベースとしたモデルとXGBoostを比較した論文を解説。
DNNとXGBoostの両方を用いたアンサンブル学習が良い性能が出たという実験結果などを紹介します。
Shwartz-Ziv, Ravid, and Amitai Armon. "Tabular Data: Deep Learning is Not All You Need." arXiv preprint arXiv:2106.03253 (2021).
* Satoshi Hara and Kohei Hayashi. Making Tree Ensembles Interpretable: A Bayesian Model Selection Approach. AISTATS'18 (to appear).
arXiv ver.: https://arxiv.org/abs/1606.09066#
* GitHub
https://github.com/sato9hara/defragTrees
表形式データのために提案されたDNNをベースとしたモデルとXGBoostを比較した論文を解説。
DNNとXGBoostの両方を用いたアンサンブル学習が良い性能が出たという実験結果などを紹介します。
Shwartz-Ziv, Ravid, and Amitai Armon. "Tabular Data: Deep Learning is Not All You Need." arXiv preprint arXiv:2106.03253 (2021).
* Satoshi Hara and Kohei Hayashi. Making Tree Ensembles Interpretable: A Bayesian Model Selection Approach. AISTATS'18 (to appear).
arXiv ver.: https://arxiv.org/abs/1606.09066#
* GitHub
https://github.com/sato9hara/defragTrees
This document summarizes a presentation on Bayesian deep learning and probabilistic programming. It discusses:
1. The history of Bayesian neural networks from 1987 to present, focusing on key papers.
2. An overview of Bayesian deep learning methodology, including Bayesian inference, variational inference, and Monte Carlo methods.
3. Probabilistic programming libraries like Edward that combine probabilistic modeling with deep learning frameworks like TensorFlow.
4. Examples of using Edward to build Bayesian neural networks and variational autoencoders for classification and generation.
5. References on Bayesian deep learning and the use of variational inference methods like Box's algorithm.
The document discusses Python programming and data science tools like NumPy, Scikit-learn, and Cython. It provides examples of using NumPy to quickly sum a large array and speed up a prime number calculation with Cython. It also briefly mentions past Python conference talks and techniques like spectral clustering and activation functions.
1) The document introduces three recent papers on addressing bias in information retrieval.
2) The first paper proposes a method to address temporal location bias in video moment retrieval by disentangling moment representations and applying causal intervention.
3) The second paper addresses popularity bias in recommendations by decoupling popularity from matching and adjusting popularity in inference.
4) The third paper addresses selection bias in estimating user retention by applying inverse propensity weighting to click-through rate predictions.
The document proposes two new online evaluation methods called EPI-RCT and CBI-IPS to estimate the causal effect of recommendations more efficiently than traditional A/B testing. EPI-RCT uses equal probability interleaving to generate recommendation lists and estimates causal effect similar to a randomized controlled trial. CBI-IPS uses causal balanced interleaving satisfying positivity and estimates causal effect using inverse propensity scoring. Simulated experiments show that while A/B testing, EPI-RCT, and CBI-IPS provide unbiased estimates, EPI-RCT and CBI-IPS require fewer users to achieve the same level of accuracy compared to A/B testing.
Recsys2018 item recommendation on monotonic behavior chainsMasahiro Sato
This document summarizes a research paper on item recommendation using multiple types of user feedback modeled as monotonic behavior chains. It proposes a model called chainRec that uses tensor factorization and a parametric rectifier to learn embeddings for different interaction types. The model is optimized using edgewise optimization focusing on edges between consecutive interaction stages. The paper compares chainRec to ablation models on several datasets, finding it performs best, especially on recommending later interactions. Visualizations show it learns meaningful embeddings separating genres by interaction type.
3. 0: 参考文献
オススメスライド
• The Recommender Problem Revisited
• Interactive Recommender Systems
• Cross-Domain Recommender Systems
• Mining Social Networks for Recommendation
• Learning to rank for recommender systems
• Web Personalization and Recommender Systems
• Frontiers in E-commerce personalization
• Past, present, and future of Recommender Systems: an
industry perspective
• Lessons Learnt at Building Recommendation Services in
Industry Scale
25. 17: ウェブ検索におけるコラボと推薦
• 検索結果の個人化
– 短期的な関心と長期的なプロファイルを反映
– 検索クエリの曖昧性解消
– 検索結果を協調フィルタリング的方法でランキング
• コラボで検索
– 後ろから検索に口出ししたり
– 検索結果をシェアしたり
– これらをサポートするシステム(非同期や遠隔なども)
Chapter 17:Collaboration, Reputation and Recommender Systems in Social Web Search, Barry Smyth, et al.