Connecting to resource 'http://www.cs.unm.edu/~terran' ===== Resource content ===== 1: 2: 3: 4: Terran Lane 5: 6: 7: 8: 16: 17: 18: 19: 27: 30: 31: 32:
20: 21: The University of New Mexico 23: 24: 25:

Terran Lane

26: 		Great Literature of the English Language 29:
33: 34: Assistant Professor
35: 36:

37:
38: The University of New Mexico
39: Department of Computer Science
40: Farris Engineering Bldg. 325
41: Albuquerque, NM 87131-1386
42: 505-277-9609 (phone)
43: 505-277-6927 (fax)
44: terran at 48: cs.unm.edu
49: 50:

As you can see from the above, I'm an assistant professor of computer 51: science at UNM. My primary (academic) interests are in machine learning; 52: reinforcement learning, behavior, and control; and artificial intelligence 53: in general. I'm also interested in computer/information security/privacy 54: and in bioinformatics. As you can probably also tell from the layout of 55: this page, I'm not an expert on web layout or graphical design. I can discuss 56: the statistical and graph-theoretic properties of the web with you, though, 57: if you like.

58:
59:

Hiring

60:

61: The postdoctoral researcher position has been filled. Thank 62: you to all who applied. 63:

64:

CLASSES

65: 66: 80: 81:
82:

Schedule

83: My schedule for F'05 is coming soon. No, really, I promise... 84: 87: 88:
89:

RESEARCH

90: 91:

Machine learning is simultaneously a pragmatic discipline, concerned 92: with the analysis of complex data from a variety of fields, and a theoretical 93: one, concerned with the principles of what is learnable, how to represent 94: acquired knowledge, how to deal with complexity/dimensionality, the interactions 95: between learned knowledge and behavior, how to measure acquired knowledge, 96: and so on. The tools we use include statistics, algorithms, knowledge representation, 97: database theory, linear algebra, and (in recent developments) topology. 98: My personal research interests include behavioral modeling and learning to 99: act/behave (i.e., reinforcement learning), scalability, representation, and 100: the tradeoff between stochastic and deterministic modeling. All of these 101: represent different facets of my overall interest in scaling learning methods 102: to large, complex spaces and using them to learn to perform lengthy, complicated 103: tasks and to generalize over behaviors. While I attempt to understand the 104: core learning issues involved, I often situate my work in domain studies 105: in practical (well, ok, semi-practical anyway) problems. Doing so both elucidates 106: important issues and problems for the learning community and provides useful 107: techniques to other disciplines.

108:

109:
110:
Publications
111:
Scaling Techniques for Planning 112: and Learning in MDPs and POMDPs
113:
The reinforcement learning paradigm rests on the foundation of the theory 114: of Markov decision processes (MDPs) and their bigger, badder cousins, partially 115: observable MDPs (POMDPs). While tractable methods for optimal planning in 116: small MDPs have been understood for decades now, we still hit a wall when 117: we try to scale to larger domains. In this project, I'm working on techniques 118: for performing approximate planning and learning in large (e.g., 2^500 states 119: or more) models.
120:
Anomaly detection for computer 121: security
122:
A number of critical problems in computer security can be viewed as 123: distinguishing some "normal" circumstance from "anomalous" or "abnormal" 124: circumstances. For example, we can think of computer viruses as being (syntactic 125: and begavioral) abnormal modifications to normal programs.
126:
Computational modeling of RNAi
127: 128:
RNA interference is a recently discovered biological mechanism 129: that appears to be a widespread and highly evolutionarily conserved 130: (i.e., ancient) genetic immune mechanism. Research in the past five 131: years or so has shown that it is involved in defense against some 132: classes of viruses and transposons, as well as in certain cellular 133: regulatory mechanisms. The exciting feature of this mechanism is that 134: it can be exploited mechanistically to target some 135: viruses, offering hints of the first possible direct treatment for 136: viral infections, as well as to selectively knock down the expression of 137: specific genes (via posttranscriptional disruption of the 138: corresponding mRNA), greatly simplifying gene function studies. 139:

140: Unfortunately, while a reasonable qualitative picture of the mechanics 141: of RNAi has emerged, we are still far from a quantitative and 142: predictive understanding. Currently, activating sequences (siRNA or 143: dsRNA) are hand-picked employing rough "rules of thumb". Our group is 144: attempting to build more quantitative and predictive models by 145: applying machine learning-based bioinformatic techniques to genome and 146: RNAi data sets. Our goals are to produce high-accuracy predictions of 147: the activity of specific sequences and, hopefully, to shed light on 148: some of the mechanical and evolutionary details of RNAi. Along the 149: way, we hope to answer pragmatic questions such as the expected false 150: positive rate (i.e., rate of knockdown of untargeted genes) and 151: minimal covering sets for gene families. 152:

153:
154: 155:

Students

156:

157: I'm fortunate enough to have the chance to work with a number of 158: excellent students here at UNM who're deeply involved in various 159: interesting ML projects including analysis of fMRI (neuroimaging) 160: data, bioinformatics, and reinforcement learning. Check in with each 161: of them to find out what they're up to! 162:

186:

Alumni

187:

188: Here are some of the students who have graduated from my lab and gone 189: on to (hopefully illustrious) endeavors elsewhere:

190: 194: 195:
196:

Other (Academic) Activities

197: 198:

199: 200: 201: 202: 206: 209: 210: 211:
203: Center for Evolutionary and 205: Theoretical Immunology (CETI)I am a member of the UNM Center for Evolutionary and Theoretical 207: Immunology (CETI). 208:
212:

213:

Machine Learning Reading Group: The Machine 215: Learning Reading Group will meet Fri 3:30-5:00 during the Summer, 216: 2004 semester.

217:

RNA Interference Reading Group/Journal Club: The RNAi Reading Group will 219: meet on alternate Wednesdays from 1:00-2:30 during the Summer, 2004 220: semester. 221:

I was the Proceedings Chair for the Twentieth International 223: Conference on Machine Learning (ICML-2003) held in Washington D.C., 224: August 21-24, 2003.

225:

226: 227: 228: 229: 235: 243: 244: 245:
230: 231: Workshop logo image. 233: 234: 236: I was also a co-chair of a workshop held in conjunction with 237: ICML-2003 on "Machine 238: Learning Technologies for Autonomous Space Applications". The 239: workshop was excellent, and we all learned a lot about the various 240: open problems, practical issues, and directions for future work. A 241: summary of the day should be available online soon. 242:
246:

247:

I am the faculty advisor for the 248: UNM Student ACM Chapter. This chapter 249: has had a somewhat rocky history, but we're working to really get it off 250: the ground this year! 258:

259:
Last Modified on Monday, 24-Jan-2005 13:03:52 MST
260: 261: ===== EOF ======