Skip to content

Commit

Permalink
🐛 Fix expanders default
Browse files Browse the repository at this point in the history
  • Loading branch information
@kozodoi
kozodoi committed Apr 1, 2024
1 parent ee7769a commit 07d365e
Showing 1 changed file with 24 additions and 24 deletions.
48 changes: 24 additions & 24 deletions _pages/p2_talks.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -52,10 +52,10 @@ permalink: /talks/
</li>
</ul>
<ul class="no-bullets">
<span id="ndots2"></span>
<li><span id="nabs2"><p></p><b>Abstract:</b> Generative AI is rapidly spreading across different industries. Retrieval Augmented Generation (RAG) is one of the most popular applications of generative AI and LLMs. In this talk, we will explain how to build conversational assistants that leverage RAG to tap into your company’s data and provide accurate answers on user questions. We will walk through the main components of RAG systems and discuss important techniques to improve and evaluate their quality.</span></li>
<span id="dots10"></span>
<li><span id="abs10" style="display: none;"><p></p><b>Abstract:</b> Generative AI is rapidly spreading across different industries. Retrieval Augmented Generation (RAG) is one of the most popular applications of generative AI and LLMs. In this talk, we will explain how to build conversational assistants that leverage RAG to tap into your company’s data and provide accurate answers on user questions. We will walk through the main components of RAG systems and discuss important techniques to improve and evaluate their quality.</span></li>
<li>
<button class="btn" onclick="show_abstract(dots_id = 'ndots2', abs_id = 'nabs2', btn_id = 'nb2')" id="nb2">&#128220; Abstract</button>
<button class="btn" onclick="show_abstract(dots_id = 'dots10', abs_id = 'abs10', btn_id = 'b10')" id="b10">&#128220; Abstract</button>
<button class="btn" onclick="window.open('https://kozodoi.me/talks/SSAI_2024.pdf')" type="button">&#128202; Slides</button>
</li>
</ul>
Expand All @@ -68,10 +68,10 @@ permalink: /talks/
</li>
</ul>
<ul class="no-bullets">
<span id="ndots1"></span>
<li><span id="nabs1"><p></p><b>Abstract:</b> Retrieval Augmented Generation (RAG) is one of the most popular applications of generative AI and LLMs. Enhancing retrievers is key to improve quality of RAG conversational assistants. In this talk, I will walk you through the main components of RAG systems and dive intro five practical techniques to improve the retrieval quality. </span></li>
<span id="dots9"></span>
<li><span id="abs9"><p></p><b>Abstract:</b> Retrieval Augmented Generation (RAG) is one of the most popular applications of generative AI and LLMs. Enhancing retrievers is key to improve quality of RAG conversational assistants. In this talk, I will walk you through the main components of RAG systems and dive intro five practical techniques to improve the retrieval quality. </span></li>
<li>
<button class="btn" onclick="show_abstract(dots_id = 'ndots1', abs_id = 'nabs1', btn_id = 'nb1')" id="nb1">&#128220; Abstract</button>
<button class="btn" onclick="show_abstract(dots_id = 'dots9', abs_id = 'abs9', btn_id = 'b9')" id="b9">&#128220; Abstract</button>
<button class="btn" onclick="window.open('https://kozodoi.me/talks/DT_2024.pdf')" type="button">&#128202; Slides</button>
<button class="btn" onclick="window.open('https://www.youtube.com/live/xPYmClWk5O8?si=Xt867cVQNntqd58A&t=83')" type="button">&#128249; Video</button>
</li>
Expand Down Expand Up @@ -118,10 +118,10 @@ permalink: /talks/
</li>
</ul>
<ul class="no-bullets">
<span id="dots1"></span>
<li><span id="abs1"><p></p><b>Abstract:</b> Machine learning is widely used to support decision-making in financial institutions. Credit scorecards are a prominent example. Such models are trained over the labeled data of previously accepted applicants, whose repayment behavior has been observed and ignore the rejected applicants. This creates sampling bias: the training data represent a limited region of the distribution on which the model is deployed for screening new customers. In this talk, I will illustrate the adverse impact of sampling bias on training and evaluation of scoring models. I will also overview possible methods to address this problem.</span></li>
<span id="dots6"></span>
<li><span id="abs6"><p></p><b>Abstract:</b> Machine learning is widely used to support decision-making in financial institutions. Credit scorecards are a prominent example. Such models are trained over the labeled data of previously accepted applicants, whose repayment behavior has been observed and ignore the rejected applicants. This creates sampling bias: the training data represent a limited region of the distribution on which the model is deployed for screening new customers. In this talk, I will illustrate the adverse impact of sampling bias on training and evaluation of scoring models. I will also overview possible methods to address this problem.</span></li>
<li>
<button class="btn" onclick="show_abstract(dots_id = 'dots1', abs_id = 'abs1', btn_id = 'b1')" id="b1">&#128220; Abstract</button>
<button class="btn" onclick="show_abstract(dots_id = 'dots6', abs_id = 'abs6', btn_id = 'b6')" id="b6">&#128220; Abstract</button>
<button class="btn" onclick="window.open('https://kozodoi.me/talks/BMAM_2021.pdf')" type="button">&#128202; Slides</button>
<button class="btn" onclick="window.open('https://youtu.be/S174yaYEifo')" type="button">&#128249; Video</button>
</li>
Expand All @@ -145,10 +145,10 @@ permalink: /talks/
</li>
</ul>
<ul class="no-bullets">
<span id="dots9"></span>
<li><span id="abs9"><p></p><b>Abstract:</b> Demand forecasting is a prominent business use case that allows retailers to optimize inventory planning, logistics, and core business decisions. One of the key challenges in demand forecasting is accounting for relationships and interactions between articles. Most modern forecasting approaches provide independent article-level predictions that do not consider the impact of related articles. Recent research has attempted addressing this challenge using Graph Neural Networks (GNNs) and showed promising results. This paper builds on previous research on GNNs and makes two contributions. First, we integrate a GNN encoder into a state-of-the-art DeepAR model. The combined model produces probabilistic forecasts, which are crucial for decision-making under uncertainty. Second, we propose to build graphs using article attribute similarity, which avoids reliance on a pre-defined graph structure. Experiments on three real-world datasets show that the proposed approach consistently outperforms non-graph benchmarks. We also show that our approach produces article embeddings that encode article similarity and demand dynamics and are useful for other downstream business tasks beyond forecasting.</span></li>
<span id="dots5"></span>
<li><span id="abs5"><p></p><b>Abstract:</b> Demand forecasting is a prominent business use case that allows retailers to optimize inventory planning, logistics, and core business decisions. One of the key challenges in demand forecasting is accounting for relationships and interactions between articles. Most modern forecasting approaches provide independent article-level predictions that do not consider the impact of related articles. Recent research has attempted addressing this challenge using Graph Neural Networks (GNNs) and showed promising results. This paper builds on previous research on GNNs and makes two contributions. First, we integrate a GNN encoder into a state-of-the-art DeepAR model. The combined model produces probabilistic forecasts, which are crucial for decision-making under uncertainty. Second, we propose to build graphs using article attribute similarity, which avoids reliance on a pre-defined graph structure. Experiments on three real-world datasets show that the proposed approach consistently outperforms non-graph benchmarks. We also show that our approach produces article embeddings that encode article similarity and demand dynamics and are useful for other downstream business tasks beyond forecasting.</span></li>
<li>
<button class="btn" onclick="show_abstract(dots_id = 'dots9', abs_id = 'abs9', btn_id = 'b9')" id="b9">&#128220; Abstract</button>
<button class="btn" onclick="show_abstract(dots_id = 'dots5', abs_id = 'abs5', btn_id = 'b5')" id="b5">&#128220; Abstract</button>
<button class="btn" onclick="window.open('https://kozodoi.me/talks/ECML_2023.pdf')" type="button">&#128202; Slides</button>
</li>
</ul>
Expand All @@ -161,10 +161,10 @@ permalink: /talks/
</li>
</ul>
<ul class="no-bullets">
<span id="dots2"></span>
<li><span id="abs2"><p></p><b>Abstract:</b> Credit scoring models are trained on data of previously accepted credit applications, where the borrowers' repayment behavior has been observed. This creates sampling bias: the training data is not representative of the general population of borrowers. The sampling bias deteriorates the model's performance when the model is used to screen new applications. We use active learning (AL) to develop a dynamic bias correction framework. When screening incoming credit applications, AL algorithm identifies unlabeled cases (i.e., applications rejected by a scorecard) that should be labeled (i.e., granted a loan) based on the expected impact on the scoring model. Issuing a loan to risky customers incurs additional cost for the lender. At the same time, augmenting the training data with applications from the unexplored distribution regions reduces sampling bias and improves the scorecard’s performance on future credit applications. We perform an empirical study on synthetic and real-world credit scoring data to test the suggested AL-driven reject inference framework and investigate the trade-off between the labeling costs and performance gains. Preliminary results indicate that Al can improve scorecard performance and profitability.</span></li>
<span id="dots4"></span>
<li><span id="abs4"><p></p><b>Abstract:</b> Credit scoring models are trained on data of previously accepted credit applications, where the borrowers' repayment behavior has been observed. This creates sampling bias: the training data is not representative of the general population of borrowers. The sampling bias deteriorates the model's performance when the model is used to screen new applications. We use active learning (AL) to develop a dynamic bias correction framework. When screening incoming credit applications, AL algorithm identifies unlabeled cases (i.e., applications rejected by a scorecard) that should be labeled (i.e., granted a loan) based on the expected impact on the scoring model. Issuing a loan to risky customers incurs additional cost for the lender. At the same time, augmenting the training data with applications from the unexplored distribution regions reduces sampling bias and improves the scorecard’s performance on future credit applications. We perform an empirical study on synthetic and real-world credit scoring data to test the suggested AL-driven reject inference framework and investigate the trade-off between the labeling costs and performance gains. Preliminary results indicate that Al can improve scorecard performance and profitability.</span></li>
<li>
<button class="btn" onclick="show_abstract(dots_id = 'dots2', abs_id = 'abs2', btn_id = 'b2')" id="b2">&#128220; Abstract</button>
<button class="btn" onclick="show_abstract(dots_id = 'dots4', abs_id = 'abs4', btn_id = 'b4')" id="b4">&#128220; Abstract</button>
<button class="btn" onclick="window.open('https://kozodoi.me/talks/CRC_2021.pdf')" type="button">&#128202; Slides</button>
<button class="btn" onclick="window.open('https://youtu.be/sUM4B1YqkNk')" type="button">&#128249; Video</button>
</li>
Expand Down Expand Up @@ -196,10 +196,10 @@ permalink: /talks/
</li>
</ul>
<ul class="no-bullets">
<span id="dots4"></span>
<li><span id="abs4"><p></p><b>Abstract:</b> Credit scoring models support loan approval decisions in the financial services industry. Lenders train these models on data from previously granted credit applications, where the borrowers’ repayment behavior has been observed. This approach creates sample bias. The scoring model is trained on accepted cases only. Applying the model to screen applications from the population of all borrowers degrades its performance. Reject inference comprises techniques to overcome sampling bias through assigning labels to rejected cases. This paper makes two contributions. First, we propose a self-learning framework for reject inference. The framework is geared toward real-world credit scoring requirements through considering distinct training regimes for labeling and model training. Second, we introduce a new measure to assess the effectiveness of reject inference strategies. Our measure leverages domain knowledge to avoid artificial labeling of rejected cases during evaluation. We demonstrate this approach to offer a robust and operational assessment of reject inference. Experiments on a real-world credit scoring data set confirm the superiority of the suggested self-learning framework over previous reject inference strategies. We also find strong evidence in favor of the proposed evaluation measure assessing reject inference strategies more reliably, raising the performance of the eventual scoring model.</span></li>
<span id="dots2"></span>
<li><span id="abs2"><p></p><b>Abstract:</b> Credit scoring models support loan approval decisions in the financial services industry. Lenders train these models on data from previously granted credit applications, where the borrowers’ repayment behavior has been observed. This approach creates sample bias. The scoring model is trained on accepted cases only. Applying the model to screen applications from the population of all borrowers degrades its performance. Reject inference comprises techniques to overcome sampling bias through assigning labels to rejected cases. This paper makes two contributions. First, we propose a self-learning framework for reject inference. The framework is geared toward real-world credit scoring requirements through considering distinct training regimes for labeling and model training. Second, we introduce a new measure to assess the effectiveness of reject inference strategies. Our measure leverages domain knowledge to avoid artificial labeling of rejected cases during evaluation. We demonstrate this approach to offer a robust and operational assessment of reject inference. Experiments on a real-world credit scoring data set confirm the superiority of the suggested self-learning framework over previous reject inference strategies. We also find strong evidence in favor of the proposed evaluation measure assessing reject inference strategies more reliably, raising the performance of the eventual scoring model.</span></li>
<li>
<button class="btn" onclick="show_abstract(dots_id = 'dots4', abs_id = 'abs4', btn_id = 'b4')" id="b4">&#128220; Abstract</button>
<button class="btn" onclick="show_abstract(dots_id = 'dots2', abs_id = 'abs2', btn_id = 'b2')" id="b2">&#128220; Abstract</button>
<button class="btn" onclick="window.open('https://arxiv.org/pdf/1909.06108.pdf')" type="button">&#128214; Paper</button>
<button class="btn" onclick="window.open('https://kozodoi.me/talks/ECML_2019.pdf')" type="button">&#128202; Slides</button>
<button class="btn" onclick="window.open('https://kozodoi.me/talks/ECML_2019_poster.pdf')" type="button">&#129703; Poster</button>
Expand All @@ -214,10 +214,10 @@ permalink: /talks/
</li>
</ul>
<ul class="no-bullets">
<span id="dots5"></span>
<li><span id="abs5"><p></p><b>Abstract:</b> In credit scoring, feature selection aims at removing irrelevant data to improve the performance of the scorecard and its interpretability. Standard feature selection techniques are based on statistical criteria such as correlation. Recent studies suggest that using profit-based indicators for model evaluation may improve the quality of scoring models for businesses. We extend the use of profit measures to feature selection and develop a wrapper-based framework that uses the Expected Maximum Profit measure (EMP) as a fitness function. Experiments on multiple credit scoring data sets provide evidence that EMP-maximizing feature selection helps to develop scorecards that yield a higher expected profit compared to conventional feature selection strategies.</span></li>
<span id="dots1"></span>
<li><span id="abs1"><p></p><b>Abstract:</b> In credit scoring, feature selection aims at removing irrelevant data to improve the performance of the scorecard and its interpretability. Standard feature selection techniques are based on statistical criteria such as correlation. Recent studies suggest that using profit-based indicators for model evaluation may improve the quality of scoring models for businesses. We extend the use of profit measures to feature selection and develop a wrapper-based framework that uses the Expected Maximum Profit measure (EMP) as a fitness function. Experiments on multiple credit scoring data sets provide evidence that EMP-maximizing feature selection helps to develop scorecards that yield a higher expected profit compared to conventional feature selection strategies.</span></li>
<li>
<button class="btn" onclick="show_abstract(dots_id = 'dots5', abs_id = 'abs5', btn_id = 'b5')" id="b5">&#128220; Abstract</button>
<button class="btn" onclick="show_abstract(dots_id = 'dots1', abs_id = 'abs1', btn_id = 'b1')" id="b1">&#128220; Abstract</button>
<button class="btn" onclick="window.open('https://www.researchgate.net/publication/335485098_Profit-Oriented_Feature_Selection_in_Credit_Scoring_Applications')" type="button">&#128214; Paper</button>
<button class="btn" onclick="window.open('https://kozodoi.me/talks/OR_2018.pdf')" type="button">&#128202; Slides</button>
</li>
Expand All @@ -242,10 +242,10 @@ permalink: /talks/
</li>
</ul>
<ul class="no-bullets">
<span id="dots6"></span>
<li><span id="abs6"><p></p><b>Abstract:</b> The talk provides a brief practical-driven introduction into using Conda for package and environment management in Python-based data science projects. The talks covers the basics of package and environment management and provides a tutorial on the basic Conda functionality.</span></li>
<span id="dots0"></span>
<li><span id="abs0" style="display: none;"><p></p><b>Abstract:</b> The talk provides a brief practical-driven introduction into using Conda for package and environment management in Python-based data science projects. The talks covers the basics of package and environment management and provides a tutorial on the basic Conda functionality.</span></li>
<li>
<button class="btn" onclick="show_abstract(dots_id = 'dots6', abs_id = 'abs6', btn_id = 'b6')" id="b6">&#128220; Abstract</button>
<button class="btn" onclick="show_abstract(dots_id = 'dots0', abs_id = 'abs0', btn_id = 'b0')" id="b0">&#128220; Abstract</button>
<button class="btn" onclick="window.open('https://kozodoi.me/talks/APA_2021.pdf')" type="button">&#128202; Slides</button>
<button class="btn" onclick="window.open('https://youtu.be/jraqbUMc4EU')" type="button">&#128249; Video</button>
</li>
Expand Down

0 comments on commit 07d365e

Please sign in to comment.