Computing Research Association honors two Khoury undergrads for their research
计算机研究协会表彰两名科里大学本科生的研究成果
February 26, 2020
2020年2月26日
By Ysabelle Kempe
伊莎贝尔·肯普
“It’s a big deal,” says Ben Hescott, Khoury College Associate Dean of Students.
“这是一件大事，”科里学院学生副院长本·赫斯科特说。
He is referring to the recent recognition of two Khoury College students as Outstanding Undergraduate Researchers by the Computing Research Association (CRA). Jeffrey Champion (BS, Math and Computer Science, ‘20) was named as a finalist, and Eric Lehman (BS, Computer Science ‘20) was awarded an honorable mention.
他指的是计算机研究协会（CRA）最近承认两名霍利学院学生为杰出的本科生研究员。Jeffrey Champion（数学与计算机科学学士，20岁）被提名为决赛选手，Eric Lehman（计算机科学学士，20岁）被授予荣誉奖。
“The CRA is choosing students that are not only doing high-quality research, but clearly are making a contribution to that research area,” Hescott says. “These are creative, thoughtful, innovative ideas, and they’re producing work in a mentored, yet independent, way.”
赫斯科特说：“CRA选择的学生不仅是做高质量研究的学生，而且显然是在为这一研究领域做出贡献的学生。”。“这些都是创造性的、深思熟虑的、创新的想法，它们以一种指导性的、独立的方式产生作品。”
Champion’s main research has been at the intersection of secure multiparty computation and differential privacy. He conducted it during his time as a research co-op with Northeastern University’s Cybersecurity and Privacy Institute. Champion says his work is best described with a concrete example.
Champion的主要研究方向是安全多方计算和差异隐私的交叉点。他在东北大学网络安全与隐私研究所（Northeast University's Cybersecurity and Privacy Institute）担任研究合作伙伴期间进行了这项研究。冠 军说他的工作最好用一个具体的例子来描述。
Suppose you have two hospitals, Champion says, that want to share aggregate statistics, but can’t disclose private patient data to the other. The process can be completed with secure multiparty computation, a cryptographic protocol that allows parties to “compute a function such that no party learns more than their own input and the output of the function,” according to Champion. But there’s a catch with that solution.
Champion说，假设你有两家医院想要共享汇总统计数据，但不能向另一家披露私人病人数据。Champion称，这个过程可以通过安全的多方计算来完成，这是一个密码协议，允许各方“计算一个函数，使任何一方只学习自己的函数输入和输出”。但这个解决方案有一个缺陷。
“It’s been pretty well-studied that the output of some aggregate function can leak information about the individual people in the dataset,” Champion says. “What’s commonly used is differential privacy, which takes an original function and incorporates randomness in certain parts. If this randomness is generated correctly, your output will be so-called ‘private.’”
Champion说：“我们已经很好地研究过，一些聚合函数的输出会泄露数据集中个人的信息。”。“常用的是差异隐私，它具有原始功能，在某些部分包含随机性。如果正确生成这种随机性，您的输出将被称为“私有”
This randomness makes the function more difficult to compute for either participating party. The purpose of Champion’s research is to develop more efficient methods to securely run differentially private algorithms on large datasets. For this particular project, Champion improved the procedure of generating many samples of random noise in a secure computation. The result isn’t just applicable in the previous hospital example; it is relevant in any process involving large, private multi-party computations.
这种随机性使得参与方的函数更难计算。Champion研究的目的是开发更有效的方法，在大型数据集上安全地运行差异私有算法。对于这个特定的项目，Champion改进了在安全计算中生成许多随机噪声样本的过程。这个结果不仅适用于以前的医院示例；它与涉及大型、私有多方计算的任何过程都相关。
“For me, this was an introduction to how good research is conducted,” Champion says.
“对我来说，这是一个介绍如何进行良好的研究，”冠 军说。
Lehman, the other Northeastern student honored, conducted research that, like Champion’s, can be directly applied to the medical field. His work, however, focuses on how to better equip doctors with the information needed to prescribe patients the best drug or intervention for their condition, especially in regards to cancer treatment.
另一位获得东北大学荣誉的学生雷曼（Lehman）进行了一项研究，与Champion的研究一样，可以直接应用于医学领域。然而，他的工作重点是如何更好地为医生提供所需的信息，以便为病人开最适合他们病情的药物或干预措施，特别是在癌症治疗方面。
His research uses a natural-language processing model to parse a medical trial report, in order to determine what it says about an intervention’s efficacy with respect to a given outcome. For example, does an article provide evidence supporting the use of aspirin to reduce risk of stroke compared to a placebo?
他的研究使用自然语言处理模型来分析一份医学试验报告，以确定报告中对某一干预措施对给定结果的有效性的描述。例如，一篇文章是否提供了与安慰剂相比使用阿司匹林降低中风风险的证据？
Additionally, Lehman’s team hired doctors to mark important language in sample trials. This data allows the researchers to train machine-learning models to recognize where in the text the answer lies, not simply spit out a conclusion.
此外，雷曼兄弟的团队还聘请了医生，在样本试验中标记重要的语言。这些数据使研究人员能够训练机器学习模型，以识别答案在文本中的位置，而不是简单地给出结论。
Lehman has also conducted research that uses machine learning to increase the accuracy of devices that detect heart arrhythmias, as well as worked with the nonprofit Cures Within Reach for Cancer, helping the organization better understand how certain interventions affect cancer survival and more.
雷曼兄弟还进行了一项研究，利用机器学习提高检测心律失常设备的准确性，并与非营利机构癌症治疗中心合作，帮助该组织更好地了解某些干预措施如何影响癌症生存等等。
“I like to do research in the medical field because these things actually have uses for people,” Lehman says. “I like the idea that what I’m working on will have a tangible impact.”
“我喜欢在医学领域做研究，因为这些东西实际上对人有用处，”雷曼说。“我喜欢这样的想法，我正在做的工作将产生切实的影响。”
Both Champion and Lehman plan to pursue PhDs, a goal typical of those honored by the CRA, according to Hescott. He points to the recognition as a beacon of light pointing toward these students’ futures.
根据赫斯科特的说法，冠 军和雷曼都计划攻读博士学位，这是CRA所推崇的典型目标。他指出，这种认识是指向这些学生未来的灯塔。
“Schools these students are applying to recognize the CRA,” Hescott says. “For Northeastern’s Khoury College to have two of our students in this space is fantastic.”
赫斯科特说：“这些学生申请的学校是为了承认CRA。”。“对于东北大学的Khoury学院来说，让我们的两个学生在这个空间里学习真是太棒了。”
如需进一步了解，或有任何相关疑问，欢迎在线咨询留学专家。如果您对自己是否适合留学还有疑虑，欢迎参与前途出国免费评估，以便给您进行准确定位
