PCPhilip K. Chan

Philip K. Chan

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Philip K. Chan has a total of 73 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Feature Decoupling in Self-supervised Representation Learning for Open Set Recognition

    by Jingyun Jia, Philip K. Chan

    Assuming unknown classes could be present during classification, the open set recognition (OSR) task aims to classify an instance into a known class or reject it as unknown. In this paper, we use a two-stage training strategy for the OSR problems. In the first stage, we introduce a self-supervised feature decoupling method that finds the content features of the input samples from the known classes. Specifically, our feature decoupling approach learns a representation that can be split into content features and transformation features. In the second stage, we fine-tune the content features with the class labels. The fine-tuned content features are then used for the OSR problems. Moreover, we consider an unsupervised OSR scenario, where we cluster the content features learned from the first stage. To measure representation quality, we introduce intra-inter ratio (IIR). Our experimental results indicate that our proposed self-supervised approach outperforms others in image and malware OSR problems. Also, our analyses indicate that IIR is correlated with OSR performance.


  • Open Access

    Representation learning with function call graph transformations for malware open set recognition

    by Jingyun Jia, Philip K. Chan

    Open set recognition (OSR) problem has been a challenge in many machine learning (ML) applications, such as security. As new/unknown malware families occur regularly, it is difficult to exhaust samples that cover all the classes for the training process in ML systems. An advanced malware classification system should classify the known classes correctly while sensitive to the unknown class. In this paper, we introduce a self-supervised pre-training approach for the OSR problem in malware classification. We propose two transformations for the function call graph (FCG) based malware representations to facilitate the pretext task. Also, we present a statistical thresholding approach to find the optimal threshold for the unknown class. Moreover, the experiment results indicate that our proposed pre-training process can improve different performances of different downstream loss functions for the OSR problem.

  • Open Access

    Self-supervised Detransformation Autoencoder for Representation Learning in Open Set Recognition

    by Jingyun Jia, Philip K. Chan

    The objective of Open set recognition (OSR) is to learn a classifier that can reject the unknown samples while classifying the known classes accurately. In this paper, we propose a self-supervision method, Detransformation Autoencoder (DTAE), for the OSR problem. This proposed method engages in learning representations that are invariant to the transformations of the input data. Experiments on several standard image datasets indicate that the pre-training process significantly improves the model performance in the OSR tasks. Meanwhile, our proposed self-supervision method achieves significant gains in detecting the unknown class and classifying the known classes. Moreover, our analysis indicates that DTAE can yield representations that contain more target class information and less transformation information than RotNet.

  • Real-world outcomes in patients with metastatic renal cell carcinoma treated with first-line nivolumab plus ipilimumab.

    by Gurjyot K. Doshi, Nicholas J. Robert, Liwei Chen, Philip K Chan, Viviana Del Tejo, Brian Stwalley, Stephen Huo, Daniel M. Geynisman

    305 Background: Nivolumab plus ipilimumab (NIVO+IPI), a first-in-class combination immunotherapy, was approved by the US Food and Drug Administration in April 2018 for the treatment of intermediate- or poor-risk advanced renal cell carcinoma (RCC), and the treatment paradigm has changed dramatically over the past few years. This real-world study examined treatment patterns and sequences, treatment response, safety, and survival outcomes with this novel first-line (1L) combination therapy among patients diagnosed with metastatic RCC (mRCC) in a community oncology practice setting. Methods: A retrospective analysis of the US Oncology Network’s iKnowMed medical data examined adult patients with an International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) prognosis of intermediate- or poor-risk clear cell mRCC who received 1L NIVO+IPI between April 1, 2018, and March 31, 2020. Baseline demographic and clinical characteristics, treatment patterns, and treatment sequence were examined descriptively. Overall survival (OS), progression-free survival (PFS), objective response rate (ORR), duration of response (DoR), time to treatment discontinuation (TTD), and time to treatment response (TTR) were analyzed using the Kaplan–Meier method. Treatment-related adverse events (TRAEs) and healthcare resource utilization were also analyzed. Results: A total of 193 adults with stage IV mRCC treated with 1L NIVO+IPI were identified. Median age (range) was 63 (30.0–89.0) years, 73.1% were male, 69.4% were white, 56.7% were IMDC intermediate risk, and 60.6% had a documented Eastern Cooperative Oncology Group performance status of 0 or 1. Median follow-up time (range) was 9.7 (0.1–24.7) months. Median PFS (95% CI) was 17.1 (12.6–21.2) months, and the 1-year landmark PFS rate was 58.3% (50.4–65.4). At 12 months, the OS rate (95% CI) was 75.4% (67.8–81.4). The ORR (95% CI) was 43.2% (34.6–52.1) among patients with a response assessment; the median TTR (range) was 2.8 (0.3–4.1) months and median DoR was not reached. Median TTD (95% CI) was 5.8 (4.5–7.5) months. Among the 63 (31.3%) patients who received second-line therapy, 50.8% received cabozantinib and 12.7% received pazopanib. TRAEs were reported in 47.2% of patients—the most frequently reported were fatigue (13.5%), rash (10.4%), diarrhea (6.7%), nausea (6.2%), colitis (3.6%), and pruritus (3.6%). The treatment-related hospitalization rate was 5.5% and the emergency department visit rate was 3.1%. Conclusions: This real-world study supports the clinical efficacy of 1L NIVO+IPI for patients with mRCC. Our findings also suggest that the NIVO+IPI combination was generally well tolerated in the real-world setting, with low rates of adverse events and healthcare resource utilization.


  • Open Access

    Engaging Undergraduate Students In Machine Learning Research: Progress, Experiences And Achievements Of Project Emd Mlr

    by Georgios Anagnostopoulos, Michael Georgiopoulos, Ken Ports, Richie Samuel, Melinda White, Veton Kepuska, Philip Chan, Annie Wu, Marcella Kysilka

  • Identifying Pros and Cons of Product Aspects Based on Customer Reviews

    by Ebad Ahmadzadeh, Philip K. Chan

    The task of identifying pros and cons from product reviews has applications in decision support for consumers. It becomes even more useful when the pros and cons are identified for product aspects so consumers can quickly see strengths and weaknesses of each aspect of the product without reading all reviews. Given a collection of product reviews, we automatically extract relevant product aspects, find the most significant sentences that represent pros and cons for each aspect, and provide a summary for each aspect. We introduce SS2 to select sentences that are likely to represent pros/cons and are semantically related to the aspect to which they are associated. Our results on three data sets indicate that compared to an existing algorithm, our algorithm can generate more meaningful summarized aspects, along with a list of pros and cons more closely related to each aspect.

  • Open Access

    Learning a Neural-network-based Representation for Open Set Recognition

    by Mehadi Hassen, Philip K. Chan

    Open set recognition problems exist in many domains. For example in security, new malware classes emerge regularly; therefore malware classification systems need to identify instances from unknown classes in addition to discriminating between known classes. In this paper we present a neural network based representation for addressing the open set recognition problem. In this representation instances from the same class are close to each other while instances from different classes are further apart, resulting in statistically significant improvement when compared to other approaches on three datasets from two different domains.